Chemical Properties of Central African Carbonado and Its Genetic Imp~Cations

Home , Carbonado

Gewhimica et CosmochimicaA cta, Vol. 58, No. 12,p p. 2629-2638, 1994 Copyright0 1994 ElsevierS cienceL td Pergamon Printedin the USA. All rightsreserved 0016-7037/94$ 6.00 + .oO

Chemical properties of Central African carbonado and its genetic imp~cations

HIROYUKI KAGI,‘* JSAZUYA TAKAHASHI,~ HIROSHI HIDAKA,~+ and AKIMASA MASUDA~$ ‘Instituteof MaterialsS cience,U niversityof Tsukuba, Tsukuba, Ibaraki3 05, Japan th Science Laboratory, Institute of Physical and Chemical Research ( RImN), Hirosawa, Wako-shi, Saitama 35 i-01, Japan SDepartment of Chemistry, Faculty of Science, University of Tokyo, Hongo, Tokyo f 13, Japan

(Received December 8, 1992; accepted in revisedform February 2, 1994)

Abstract-Carbonado diamonds from Central African Republic exhibit very intense laser-induced photoluminescence due to the presence of the radiation damages. From the spectral profiles of photoluminescence obtained for these samples, the carbonado stones were grouped into three groups depending on a variation in the structure of m~ation~amage products; Group-A carbonado with 504 nm band (3H band), Group-B carbonado with 575 nm band in addition to weak background peaks centered at 512 and 580 nm, and the Group-AB carbonado with features intermediate between Group-A and Group-B. When heating Group-A carbonado, Group-A was found to transform to Group-B at temperatures higher than 400’C. Thermogravimetry and Differential Thermal Analysis (TG-DTA) clarified the difference in thermal properties between Group-A and Group-B. While Group-A carbonado yields one intense and narrow exothermic peak at 8OO”C, Group-B carbonado exhibits a broad exothermic peak at 785X The distinct difference in the oxidation rate could be att~buted to the difference in the density of micropores in carbonado. Chondrite-normalized rare earth element ( REE ) patterns for the bulk carbonado samples show negative Eu anomalies, and the inclinations of the patterns are similar to those of kimberlites and much steeper than that of continental crust as represented by continental shales. In the chondrite-normalized value, La/ Lu ratios were of the order of lo2 and the light REE abundances were as high as IO3 relative to chondrite. Further, 1384 cm-’ absorption band was observed in IR (infrared) spectra of both Group-A and Group-B, which suggests the presence of nitrogen platelet in carbonado. These REE and IR results imply that the seed crystal of carbonado was formed at high pressure and high temperature and was subjected to the annealing process to generate nitrogen platelets. The genesis ofcentral African carbonado is discussed as a reconciliation of these experimental results.

INTRODUCf’ION lamproite (F. V. Kaminskii, pers. commun., 1991). On the basis of the above observations, large impact me~mo~hism CARB~NAD~S ARE AGGREGATESof randomly polycrystalIized on the Earth’s crust in the Precambrian era (SMITH and diamond particles with some noncarbonaceous inclusions, DAWSON, 1985) or transformations of subducted organic and their genetic history is open to debate. Their color is carbon to diamond (ROBINSONe t al., 1978) were proposed usually grayish to black or sometimes is light brown due to for their possible origins. On the other hand, KAMINSKII the presence of amorphous carbon or graphite, various me- ( 1987) speculated that carbonados were formed in the crust tallic oxides, and minerals. In addition, radiation-induced from coals upon exposure to uranium and thorium radiation damage in the diamond crystaIlite mi@ be related to the without undergoing the high pressure prevailing in the upper visual color. They have inclusions (silicates, phosphates, and mantle. Recently, OZIMA et al. ( 199 1) reported on the noble oxides) potentially related to the Earth’s crust ( FRANT~ESSON gas isotope compositions revealing the presence of sponta- and KAMINSKII, 1974; TRUEB and BUTTERMAN, 1969; TRUEB neous fission products which is consistent with the KAM~NSKII and DE WYS, 1969; MORIYOSHI et al., 1983) and contain ( 1987) speculation. Further, radiation-damage products in polycyclic aromatic hydrocarbon (PAH) ( KAMINSKII et al., carbonado were observed in laser-induced photduminesc- 198.5). Carbon isotopic studies ( VIN~RA~V et al., 1966; ence, and direct evidence of the exposure was demonstrated GALIMOV et al., 1985 ) showed that Brazilian carbonado was (&Gi et al., 199 1). The photoluminescence profiles are ex- significantly 13C-depleted and it has been suspected that such pected to give a clue to the thermal history of the carbonado ‘3C-depleted diamonds might not be of kimberlitic origins diamonds, because the optical properties of the color center ( KIRKLEY et al., 1991). (However, SMIRNOV et al. ( 1979) produced by radiation of high-energy particles are fairly sen- found S13C values as low as -20.5%0 in diamonds from Le- sitive to temperature. Thus, the close bearings of radiation sothan kimberlites.) Until now, carbonado has been found from radio-active nuclides on the formation of carbonado only in placers, and has never been found in kimberlite nor have been ascertained, but the connection to coals remains pure speculation. Apart from the specific geochemical properties of carbon- * Author to whom correspondence should be addressed. ado diamond, its superior fracture toughness to a single crystal + Present address: Department of Chemistry, Faculty of Science, of diamond is worthwhile to note from a viewpoint of material Tokyo Metropolitan University, Minami-Ohsawa, Hachioji, Tokyo 192-03, Japan. science. It is also expected that info~ation on the genesis * Presem address Department of Chemistry, University of Electro- of carbonado elucidated by the geochemical study would Communications, Chofu, Tokyo 182, Japan. throw light on the artificial sintering process of diamond. In 2629 2630 H. Kagi et al. this study, the investigation of laser-induced photolumines- and subsequent cooling history of the diamond. IR spectra ofdiamond cence was furthered, and a close link between luminescence give us the valuable information on the structural state of nitrogen aggregation which is sensitive to the thermal history of the diamond. profile, thermal properties, and REE abundances was estab- For the IR measurement, carbonado sample was crushed in stainless lished. Furthermore, in order to investigate the environment vessel to the size about 0. I mm. Prior to preparation of a KBr pellet of diamond growth, IR spectra were studied and atomic forms for the IR measurement, the crushed grains were leached with a mix- of the nitrogen contained in carbonado were deduced. Be- ture of hydrochloric acid and hydrofluoric acid at 80°C for a few hours in order to remove silicate mineral inclusions, although com- sides, REE isotope compositions were investigated in a search plete elimination of the inclusions could not be expected judging for spontaneous fission products in the REEs. The results of from the specific structure of carbonado, i.e., submicron grain size these versatile experiments provide further clues to the origin of diamond microcrystal and rigid bonding between the diamond of carbonado diamond. microcrystals. The specimens were examined in a transmission mode with a JASCO FI/IR-3 Ff-IR spectrometer fitted with an X8 beam condenser. All absorption spectra were obtained under room tem- EXPERIMENTAL METHODS perature with a repeated scanning of 1000 times and a spectral res- Samples olution of 4 cm-‘. All carbonado stones (uncrushed grains) studied here are from REE Analysis alluvial deposits in the Central African Republic, and their average grain size is 5-10 mm. Like Brazilian carbonado, the Central African Five carbonado stones unleached were chosen for REE analysis carbonado is an aggregate of polycrystallized diamond particles with so that they represent differing groups based on photoluminescence a considerable amount of noncarbonaceous inclusions. There is no spectra measurements. The bulk samples were washed in acetone geological information available about the details of occurrence of with ultrasonic at room temperature and were heated in distilled HCI the carbonado stones studied. In view of the Central African origin at 60°C for one hour, for the purpose of removing the stains on the they are alluvial. The sample surfaces are tinged dark red when viewed surface of the sample. Then, about 0.2 gram of carbonado grain was under a microscope. Acid washings from the samples were yellow, heated in a platinum crucible for half an hour at 900°C in air. Sub- which indicate carbonado contains iron oxide (hematite) as an in- sequently, the carbonado was completely ashed. (In this combustion clusion and acid leachate is colored by ferric ion. process, platinum crucible catalyzed the decomposition of diamond because no definite oxidation reaction was observed in heating with Photoluminescence Spectra a porcelain crucible around 7OO”C,w hilst heating in platinum crucible under the same temperature condition gave rise to complete ashing Laser-induced photoluminescence spectra were obtained for thirty- of carbonado. The catalytic action of platinum has been confirmed seven carbonado stones with a Jobin Yvon RAMANOR U-1000 from DTA.) Isotope measurements were made on one carbonado double monochromator at room temperature. The excitation radia- stone which belonged to Group-B (see Results) based on photolu- tion was the 488.0 nm blue line from an Arf gas laser, model l68B minescence spectra. Determination of the isotopic compositions of manufactured by Spectra Physics. The laser power was 10 mW at cerium, neodymium, samarium, and europium was made on a JEOL the sample surface, and the laser beam is focused on a spot of ap- JMS-OSRB thermal ionization mass spectrometer after separation proximately 2 pm in diameter with an objective lens. The photon from major elements followed by mutual REE separation using alpha- counting system consists of a PMT-R464S (photomultiplier) and a hydroxy-isobutyric acid (HIBA). In this measurement, no corrections PHC-C 1230s (photon counter) manufactured by Hamamatsu Pho- for isotope mass fractionation by use of standard normalizing values tonics. All spectra were obtained covering the region from 492 to were made, but there was enough accuracy and precision to recognize 614 nm in wavelength, which corresponded to 200-4200 cm-’ in the sizable fissiogenic isotope anomaly. The precise analytical pro- Raman shift. Photoluminescence spectra were taken in a scanning cedure employed here and its accuracy and precision were docu- mode with a step size of 2 cm-’ and an integrating time of 0. I second mented in HIDAKA and MASUDA( 1988). per step. The sample solution for the determination ofthe REE abundances was diluted for the purpose of controlling the REE concentration Proton Irradiation of Artificial Diamond level below 50 ppb. REE abundances were determined with VG Plasma-Quad ICP mass spectrometer. For making a correction of For comparative studies, in order to clarify the optical properties the matrix effect in the ICP ionization process, we added I ppm Cd of irradiated diamond grain, we carried out proton implantation fol- solutions to both the sample solution and the standard solution as lowed by photoluminescence measurement using an artificial dia- an internal standard. In this study, the REE pattern obtained with mond. The diamond sample we used was grown by the combustion ICP-MS was crosschecked by a stable isotope dilution method using flame method (HIROSE et al., 1990), one of the Chemical Vapor thermal ionization mass spectrometer. The coincidence of REE pat- Deposition (CVD) processes for diamond growth. The grain size of terns obtained by the two methods was confirmed. The error for the the grown crystals was of the order of IO pm and a molybdenum analytical method using ICP-MS was estimated to be within +5% plate of 0.2 mm thickness was used for the substrate. The irradiation ( HIRATAe t al., 1988) regarding the multi-isotope REEs, giving enough of the CVD diamond was done with a tandem accelerator; the total precision for the discussion of REE pattern. dose, 6.8 X lOI H+ ions (1.09 X 10m4c oulomb) at an energy of 3 MeV on a circular spot in diameter of approximately 2 mm. RESULTS Thermal Analysis Photoluminescence Spectra Thermogravimetry and Differential Thermal Analysis (TG-DTA) measurements were carried out in dry air using a Rigaku TG-DTA All carbonado diamonds studied here show very intense apparatus with a heating rate of 20°C min -I. The heating operation photoluminescence bands in the visible light region, and the was carried out in the range from 100 to 900°C. For holding sample Raman line at 1333 cm-’ (522 nm in wavelength) derived in the heating operation, a platinum pan was used. from the lattice vibration of diamond cannot be detected because the Raman intensity is much weaker than that of Infrared ( IR) Spectra the photoluminescence. Figure 1 shows typical photolumi- Most natural diamonds contain nitrogen as the main impurity. nescence spectra, and the spectral profiles can be classified Nitrogen can be contained up to 0.5 wt% and can be present in mainly into three groups from their characteristic band several states of aggregation depending on the conditions of formation shapes. The first one (Fig. la) has one strong photolumi- Radiation damage and REE profiles of carbonado diamonds, Central African Republic 263 1

strong after high-energy electron radiation ($2 MeV). The photoluminescence band due to the 3H center disappears upon annealing at 700 K in diamonds with high nitrogen content and is stable to higher temperature in type IIb (boron- doped) diamond (DAVIES, 1977; WALKER, 1979). On the other hand, the 57.5 nm band is believed to involve the single nitrogen in association with a vacancy (V + N center) (COL- LINS and LAWSON, 1989), but a divacancy (V + V center) is also considered as a possible structure (DEAN, 1965). Either way the annealing of diamond containing a radiationdam- aged product gives rise to the 575 nm band. Therefore, remarkable spectral changes can be expected after heating Group-A carbonado. One of the crushed Group-A carbonado particles of submillimeter grain size was heated at 500°C for half an hour. The resultant spectrum shown in Fig. 3a exhibits the disap- pearance of the 3H band at 504 nm, and the heated Group- A carbonado shows a quite similar spectral profile to those of Group-B carbonado displayed in Fig. 3b. This transition takes place between 400 and 500°C ( KAGI et al., 1991). The two broad backgrounds observed in both spectra can be iden- tified as the band-A photoluminescence which is attributed to radiative recombination across the indirect energy gap of electrons and holes separately trapped at donor and acceptor centers ( DEAN, 1965). Fu~ermore, in order to confirm the homogeneity of the photolumine~en~e feature within a single so0 550 640 WAVELENGTH/urn FIG. 1. Photoluminescence spectra of natural polycrystal diamonds, carbonado, excited by 488 nm blue line from Ar” gas laser. (a) Group- (a) A carbonado; (b) GroupAB carbonado (intermediate type between Group-A and Group-B); (c) Group-B carbonado.

nescence band at 504 nm with a weak adjacent background (Group-A carbonado). The second (Fig. 1c ) has one strong photolumine~nce band centered at 575 nm and two broad bands at 512 and 580 nm (Group-B carbonado). The third (Fig. I b) has the intermediate type of photoluminescence which includes both spectral features of Group-A and Group B. One can find that these observed photoluminescence bands J can be assigned to radiation-induced defects (DAVIES, 1977; WALKER, 1979), which is consistent with the presence of fissiogenic noble gases in carbonado suggesting the influence (b) 1333cm-t in Raman shift of uranium and thorium ( OZIMA et al., 199 1). The 504 nm J photoluminescence band is assigned to radiation-damaged vacancy (V), and is called as the 3H defect center with rhom- bic I symmetry (WALKER, 1975). For generating the 3H center in diamond crystals in laboratory, implantation of some accelerated high-energy particles is necessary. In Fig. 2, photoluminescence spectra of artificially irradiated CVD diamond crystals are shown. The spectral measuring conditions were the same as those for carbonado diamonds. Raman line at 1333 cm-‘, which had been observed before the irradiation I IIIII,I,II (Fig. 2b), was overcome by appearance of the strong 3H so0 550 600 photoluminescence band in the spectrum after the irradiation WAVELENGTH/run (Fig. 2a). One can find that the photoluminescence spectrum given in Fig. 2a shows the same profile as that of Group-A FIG. 2. Change in the photoluminescence spectra of artificial CVD diamond crystals with the irradiation of 3 MeV proton. (a) photo- carbonado in Fig. la. It is known that the 3H center appears Iuminescence spectrum after the irradiation; (b) photoluminescence in all types of diamond after irradiation, and is especially spectrum before the irradiation. 2632 H. Kagi et al.

thennic band at 785*C with a gradual lowering of the TG curve. One can recognize the tailing of the exothermic curve of Group-B carbonado toward the lower temperature. The experimental results prove that the oxidation of Group-B carbonado begins at much lower temperature than that of GroupA carbonado, which means that the diffusion rate of oxygen gas is much higher in Group-B than in Group-A carbonado. The remarkable difference in oxidation behavior between the two groups implies the distinction of micro structures between differently grouped carbonados, which might have resulted from differences in the formation conditions and thermal history of carbonado grains such as etching by hydrothe~~ fluid. These TG-DTA results suggest that micropores in Group-B carbonado are denser than those in GroupA carbonado.

IR Spectra

Prior to inspection of the IR data taken for our samples, it would be of some use to summarize the general information about IR spectra of diamond. Classification of diamond can be made basically by the IR spectra. In all kinds of diamond

I I iIIIII,iI 500 550 600 WAVELENGTH/inn

FIG. 3. Photoluminescence spectra of heated carbonado. (a) Group- A carbonado heated to 500°C for half an hour; (b) Group-B carbonado. stone, several carbonado stones of about 5 mm were crushed into submillimeter grains. All crushed carbonado particles derived from one identical stone revealed the identical photoluminescence profile. SpecilicaIIy, a Group-A stone com- prises exclusively Group-A particles, a Group-B stone exclu- TG 0 I I , sively Group-B particles. The positionings of the particles in 400 500 600 700 8Kl the stone (center or surface) have nothing to do with the basic characteristics of the photoluminescence. This is one TempcratureCC) of the notably interesting findings in the present work. DTA In the course of the heating experiment, we noticed dif- TG ferent behavior between Group-A and Group-B carbonado. 200 ’ (b) Group-B carbonado was completely ashed by heating at 700°C for half an hour in platinum crucible in air, while Group-A carbonado showed no change in appearance in the same heating procedure except for the foregoing photoluminescence transition from Group-A to Group-B type. These experimental results imply a difference in thermal properties between the two groups of carbonado, on which our attention is focused in the next section.

Thermal Analysis Figure 4a shows the TG-DTA curves for Group-A carbonado heated on a platinum pan in dry air. Figure 4b shows those for Group-B carbonado measured in the same condi- 800 SGU tion. Group-A carbonado gives one intense and narrow exo- Tempersture(Y?) thermic peak at 800°C associated with lowering of the TG FtG. 4. TG-DTA curves of carbonado. (a) Group-A carbonado; curve. On the other hand, Group3 gives one broad exo- (b) Group-B carbonado. Radiation damage and REE profiles of carbonado diamonds, Central African Republic 2633 either natural or artificial, intrinsic IR absorption bands are erals in carbonado. The intensity of the broad band around present in the energy region higher than 1400 cm-’ with the 1100 cm-’ is much higher than that of the defect-activated same intensity, while the defect-induced absorption bands one-phonon bands which should be detected in the energy observed below 1400 cm-’ is strongly specimen dependent range from 1100-1200 cm-‘, thus, the classification of car- (SUTHERLAND et al., 1954; LAX and BURSTEIN, 1955). Type bonado cannot be made based on the absorption bands in IIa diamonds which contain little or no nitrogen show only the range of IOOO- 1300 cm-’ . Besides, in an IR region ranging the former IR absorption. Thus, about 98% of all clear natural from 1350-1400 cm-‘, the reported inclusion in carbonado diamonds have the defect-induced one phonon lines below of minerals such as serpentine, kaolinite, quartz, rutile, il- 1400 cm-‘, and the spectral profiles fall into three groups; menite, and so on give no absorption band, so that well- type IaA, type IaB, and type Ib. Type IaA and type IaB dia- defined absorption bands in this IR region are attributable monds contain nitrogen atoms as a form of nitrogen-aggre- to diamond. As shown in Fig. 5, one sharp absorption peak gates. The nitrogen in type IaA is thought to occur mainly was observed at 1384 cm-’ both for Group-A and Group-B as nearest neighbor pairs (the ‘A’ form), and is characterized carbonado. The wavenumber of the observed band ( 1384 by the absorption bands at 1280 cm-’ and 1200 cm-’ which cm-‘) seems to be slightly higher than the wavenumber of are called A bands (SUTHERLAND et al., 1954). In type IaB the normal B2 band ( 1370 cm-‘), but this energy deviation diamonds, nitrogen is contained in a form of platelets lying can be explained by the size effect of nitrogen platelets. (It in { 100) planes of diamond lattice and B absorption bands is known that the wavenumber of the sharp band at 1370 in IR region are observed. The B bands can be subdivided cm-’ coming from nitrogen platelets in type Ia diamonds into the independent B 1 at 1175 cm-’ and B2 at about 1370 varies with nitrogen content and the size of the platelets, e.g., cm-‘, and it is concluded that the B2 band is activated by DAVIES, 1977; WALKER, 1979; CLACKSON et al., 1990.) the { 100) platelets, from the correlation between the ab- Therefore, the 1384 cm-’ band observed in this study has sorption coefficient of B2 band and the integrated X-ray spike been assigned to aggregated nitrogen and both Group-A and intensity (EVANS and PHAAL, 1962; SOBOLEV et al., 1968). Group-B carbonado can be classified into type IaB. It can be On the other hand, type Ib diamonds containing single sub- concluded that the optical property of carbonado diamonds stitutional N atoms have an intense absorption band at 1130 is the same as that of typical natural terrestrial diamonds cm-’ and an additional weak narrow line at 1350 cm-’ is from the viewpoint of the aggregation form of incorporated sometimes observed (DYER et al., 1965). nitrogen atoms in diamond lattice. On the other hand, the In Fig. 5, IR spectra for Group-A and Group-B carbonado presence of paramagnetic nitrogen has been reported are displayed in the 800- 1500 cm-’ region, in which typical ( KLYUEV et al., 1978) notwithstanding a technical difficulty IR absorption bands for various kinds of diamond are ob- of detection of such a state of nitrogen in the ESR measure- served. No remarkable difference can be observed in these ment on carbonado due to the considerable contents of para- two spectra. An intense broad band around 1100 cm-’ is magnetic metals in carbonado. (The presence of paramagnetic attributable to the absorption by silicate minerals judging metals makes it difficult to analyse the concentration of ni- from its broad band shape and the high content of clay min- trogen atoms.) One might feel uneasy about the coexistence of aggregated nitrogen platelets and substituted paramagnetic

, 1 I I I, I I I I I nitrogen atoms, because the isolated nitrogen atoms are con- sumed to form the platelets when nitrogen atoms aggregate. However, the platelet degradation was reported under the intense electron beam irradiation (BARRY, 1991). Conse- quently, we can attribute the presence of paramagnetic nitrogen atoms to the intense irradiation carbonado having suf- fered from uranium.

REE Isotope Compositions

Being aware of the large amounts of implanted Xe and Kr originating from 23sU spontaneous fission in carbonado ( OZIMA et al., 1991), fissiogenic REEs might be detected because the fission yields of the light REEs are as high as those of *6Kr and ‘36Xe (SWINDLE et al., 197 1). In Table 1, the isotope compositions of cerium, neodymium, samarium, and europium are shown both for the terrestrial standard material and carbonado. No definite devia- GmupB tions were observed between the isotope compositions of the two samples within the error bar of our measurement (2~ mean ). This means that carbonado contains a considerable amount of primordial REEs with terrestrial isotope compositions, and REEs of fission-origin cannot be observed. The origin of the REEs contained in carbonado cannot be attrib- FIG. 5. IR absorption spectrum of Group-A and Group-B carbon- uted to the fission products unlike the noble gasses in car- ado in the region from 800-1500 cm-‘. bonados. 2634 H.Kagietal.

TABLE 1. REE isotope compositions of a carbonado stone. solution. From the unique characteristics of carbonado, some scientists believe it has a crustal origin, but we believe that the crustal origin is not the adequate solution to the carbonado ce 136 0.19 f 0.01 0.19fO.01 enigma. KAMINSKII ( 1987) has speculated that carbonado 138 0.24 f 0.01 0.25 f 0.01 diamonds were formed under the influence of radiation from 140 88.30fO.10 88.40 f 0.06 142 11.26fO.10 11.16f0.05 uranium and thorium in the crust, and OZIMA et al. ( 199 1) supported his speculation from a viewpoint of noble gas geo- Nd 142 27.19 +z0 .06 27.43f0.14 chemistry. They have proposed diamond formation from or- 143 12.18f0.03 12.14f0.06 144 23.SOfO.C4 23.7710.15 ganic matters such as coal in the crust without undergoing 14.5 8.29 f: 0.02 8.26 f 0.05 high pressure and high temperature. With a motivation re- 146 17.17f0.03 17.08f0.12 148 5.74 fO.O1 5.73 f 0.04 lated with material science, nitrogen ions were implanted 150 5.62 fO.O1 5.58 f 0.06 into glassy carbon, which has a similar chemical structure to

Sm 144 3.12f0.01 3.16f0.01 that of coal ( IWAKI et al., 1990). The resultant Raman spectra 147 15.11 f0.02 15.21 f0.03 showed that the structure of the ion-implanted glassy carbon 148 11.3f0.02 11.31 *to.02 layers changed to an amo~hous-like structure with disordered 149 13.86 f 0.02 13.95 4 0.03 150 7.39 f 0.01 7.37 f 0.01 graphite. Similar Raman spectra can also be observed both 152 26.65 f0.03 26.56 f 0.05 for implanted diamond and organic polymers. Judging from 154 22.57 f0.03 22.44 f 0.04 these chemical structures of implanted carbonaceous matters, EU152 47.88 f0.06 48.00 f 0.01 it seems unlikely that the radiation energy was responsible 153 52.12f0.07 52.00 f 0.10 for the formation of diamond directly from carbonaceous matter. Thus, the most plausible product of radiation into carbonaceous materials like coals supposed as the potential REE Abundance precursor of carbonado would be amorphous carbon. Chondrite-normalised REE patterns for Central African Here an attempt is made to draw the scheme of the self- carbonado are shown in Fig. 6. For convenience, five samples consistent carbonado genesis from our experimental results. investigated for REE abundance are designated Al and A2 First, carbonado shows the REE pattern closely similar to of the Group-A, AB of the intermediate type (GroupAB), that of kimberlite; large inclination, and linearity of REE and B 1 and B2 of the GroupB, respectively. For al1 samples patterns. The high absolute content of REEs in carbonado studied, a negative europium anomaly and steep inclination was pointed out also by TRUEB and DE WYS ( 1971). They of REE pattern were observed. For all analysed samples, the chondrite-normalized lanthanum abundances are as large as 103, while the normalized lutetium abundances are of the order of 10. These REE patterns with steep gradient from lanthanum to lutetium are very similar to those of kimberlite (CULLERS and GRAF, 1984; MITCHELL, 1986). An intriguing point is the difference in REE concentration between Group IO4 A and Group-B. The variation of REE abundances among the five samples analysed is as follows: Group-A carbonado > GroupAB carbonado

> GroupB carbonado. The REE concentration, therefore, shows a good correlation with the classification given by the laser-induced photoluminescence for five samples. In addition, it is worth noting that Ho-Lu segments appear to show gentle slopes in all patterns except for sample Al. Detailed consideration of the relationship between the REE profiles and optical properties is made later.

DI!iXXJSSION Crystallization of Diamond C~st~l~tion of carbonado is the most disputable and enigmatic event in the genetic history of carbonado. The small size of crystallite and the relatively high density of dislocations I t I I I I I I I, tend to indicate a significant level of residual stress ( TRUEB la Cc R Nd SmlGuGd’IbDyHoPTmYbLu and DE WYS, 197 1; MORIYOSHIet al., 1983 ). suggesting that 6. REE the crystallites have grown under condition well above the FIG. Leedey chondrite-norrnalised patterns of carbonado. A 1 and A2 belong to GroupA carbonado, Bi and 82 to Group-B equilib~um, which resulted in rapid nucleation, a phenom- carbonado, and AB to the inte~~iate group ( GroupAB ) , respec- enon like a sudden formation of small crystals in a saturated tively. Radiation damage and REE profiles of carbonado diamonds, Central African Republic 2635 inferred a connection between carbonado and diamond- In either way, the crystallization process of diamond par- bearing alluvial sediments. AKAGI and MASUDA ( 1988) re- ticle of carbonado can be attributed to high pressure and high ported that the REE pattern of cubical diamond (cuboid 1, temperature just like typical natural diamond showing the Zaire, was very similar to that of kimberlite, and they inferred presence of aggregated nitrogen and the kimberlite-like REE that the cuboid diamond studied by them grew in the volatile- pattern. rich fluid in the kimberlite or a precursor of kimberlite. While the REE abundance level of cuboid diamond is substantially Possible Influence of Ion Implantation upon the Diamond lower than that of carbonado, the two REE patterns have Aggregate similar shape. This supports a common crystallization event Carbonado is an aggregate of diamond microcrystals. An for carbonado and cuboid diamond. The large REE abun- X-ray study on the microstructure of carbonado ( TRUEB and dance in carbonado suggests a large amount of incorporated BUTTERMAN, 1969) found no evidence for a coating or ce- inclusions as the host phase of REEs and suggests the rapid menting agent such as amorphous carbon and or silica, and crystallization process, which is consistent with the experi- concluded that the crystallites of carbonado are bonded by mental results of rapid nucleation under a growth condition interatomic forces as is usually the case in polycrystalline well above equilibrium ( WENTORF and BOVENKERK,196 1). materials. Generally speaking, it needs severe physical con- Second, the IR spectra with the sharp line at 1386 cm-’ ditions to sinter a covalent compound with a high melting suggest that carbonado was subjected to annealing under high point as is a typical case for diamond. HALL ( 1970) reported temperature to form N platelets in the diamond lattice and that the direct bonding between diamond crystallites without has an upper mantle origin. The aggregation dynamics of N a binder occurs under conditions of 1400 K and 8.5 GPa. atoms has been discussed from a viewpoint of IR spectroscopy Our experimental results revealed that Group-A carbonado (CHRENKO et al., 1977; EVANS and Ql, 1982). The large with the photoluminescence at 504 nm shows no trace of majority of natural type Ia diamonds contain nitrogen atoms annealing at a temperature higher than 700 K ( KAGI et al., in various degree of aggregated form; A center with two ni- I99 1)) whereas high temperature and high pressures are req- trogen atoms, an N3 center with three nitrogen atoms and a uisite for the artificial diamond sintering. The presence of B center with nitrogen platelets. To form nitrogen platelets the 504 nm band (3H band) in GroupA carbonado indicates starting from type Ib diamonds with singly substituted nitro- that the highest temperature experienced was lower than gen atoms, it takes the highest temperature among three types 400°C after forming the radiation-damage products in Group- of nitrogen forms: IaA, IaAB, and IaB. Thus, the aggregation A carbonado, which means GroupA carbonado diamonds of nitrogen atoms evinces that natural diamonds sojourned were irradiated under the condition of the remarkably low in the upper mantle prior to ejection to the surface of the temperature in the crust. This might seem to be inconsistent Earth (EVANS and Ql, 1982). They suggested from heating with the presence of nitrogen platelets presumed from the experiments that the type Ia diamonds spent between 200 1386 cm-’ band in IR spectra, but one can avoid the incon- and 2000 Ma in the upper mantle at temperatures between sistency by assuming that the nitrogen platelets had been 1000 and 14OO“C. Thus, the diamond microcrystals, which formed before carbonado was aggregated. Furthermore, had been source crystals for the growth of carbonado, would metamorphic overprints on diamond aggregates formed in have sojourned in the upper mantle to form nitrogen platelets. the mantle is also likely. From our versatile experimental Besides the upper mantle, the high pressure and high tem- results, we cannot decipher whether the irradiation took place perature requisite for diamond formation might be attributed before or after the aggregation of diamond microcrystals. to metamorphic rocks. SOBOLEVa nd SHATSKY ( 1990) re- Judging from the large excesses of implanted spontaneous ported that diamonds can form in situ in metamorphic mas- fission products in comparison with those estimated from sifs, and these crustal rocks may, therefore, be the source for the present uranium content (OZIMA et al., 1991), highly some alluvial microdiamonds. The microdiamonds from concentrated solutions of uranium and thorium should have metamorphic massifs are similar in crystallite size (averaging filled the grain boundaries of the microcrystals. Then, alpha 12 pm) and carbon isotopic ratio with those of carbonado, particles of 4.2 MeV must have been emitted from uranium which seems consistent with the interpretation attributing and thorium fission. According to the stopping cross-sections the origin of carbonado to the metamorphic process, but the for energetic helium ions in carbon, the 4.2 MeV alpha par- metamorphic diamond particulates are not tightly aggregated ticle has enough energy to penetrate the micron order of dia- but fragile (N. V. Sobolev, pers. commun., 1992). Conse- mond grains. It is convincing to assume that the irradiation quently, we cannot confidently ascribe the whole formation gave rise to some considerable influences on the diamond process of carbonado to the metamorphic condition. An ex- aggregate. First, it should have induced the formation of ra- perimental attempt at high pressure and high temperature diation-damage products in carbonado resulting in the pho- for the growth of polycrystalline diamond (KANDA et al., toluminescence bands at 504 nm, 575 nm, etc. Second, the 1976) showed the formation of loosely packed aggregate of irradiation would have affected the packing conditions of diamonds of about 1 pm. This result suggests that the direct diamond aggregate. Ion implantation is known to strengthen tight bonding between randomly orientated small diamond diamond. Nowadays this technique is widely used to crystals, which is the most characteristic property of carbon- strengthen materials and improve their wear resistance, either ado, requires some peculiar process other than high pressure by pinning dislocations in plastic materiels or by restricting and high temperature. A possible process to strengthen the the propagation of cracks in brittle material (WILKS and binding between diamond aggregates is discussed in the next WILKS, 1992). As mentioned above, diamond aggregates of section. randomly orientated microcrystals as primarily grown are 2636 H. Kagi et al. loosely packed both for natural and artificial origins. On the (BROOKINGS, 1984). From the above mentioned results, it other hand, the crystallites of carbonado are tightly bonded. can be assumed that carbonado had undergone intense hy- Therefore, the tight packing of microcrystals could be attrib- drothermal etching which results in leaching of uranium and uted to the irradiation of high-energy particles from uranium porous microstructure of carbonado. This assumption is and thorium. We have not investigated photoluminescence consistent with the presence of florencite in carbonado re- spectra of Brazilian carbonado, but OZIMA et al. ( 199 1) reported by TRUEB and DE WYS ( 197 1). ported the presence of fissiogenic noble gases and demon- There seem to be some distinct differences in the extent strated the exposure to uranium and thorium radiation for and strength of etching between our groupings, GroupA and Brazilian carbonado. Under these conditions, it seems to be Group-B, based on the photoluminescence. Group-B car- likely that the irradiation has close bearings on the genesis bonado has denser micropores than Group-A carbonado, of carbonado. The next problem to be considered is the phys- which is supported by the DTA results and the lower starting ical process to form direct bonding between microdiamonds temperatures of combustion for GroupB than that for Group with the high-energy irradiation. A. Further, it must be noted that the photoluminescence pro- It is known that surfaces of diamond are terminated by H files highly correlate to the density of the micropores and the atoms ( PATE, 1986; HAMZA et al., 1988; VIDALI et al., 1983; DTA results. GroupB carbonado with dense micropores VIDALI and FRANKL, 1983; WACLAWSKIet al., 1982), and would have been severely etched by the hydrothermal fluid for making direct bonding between the diamond microcrystals of higher temperature compared with that of Group-A car- it is essential to strip off the terminating hydrogen atoms. bonado because GroupB carbonado reveals the 575 nm band The high-energy alpha particles which had been emitted from suggesting exposure to high temperature (at least >4OO”C) uranium and thorium emplaced at the grain boundaries of as evinced by our experiment and the high diffusion rate of the micro crystals would give rise to the dissociation of the gases implicated by DTA results. The lower REE concentra- surface hydrogen with the energy (-MeV) of the particle. tion of GroupB carbonado compared with Group-A car- The dissociation of terminating hydrogen will cause dangling bonado can be attributed to the severe etching with the high bonds of carbon atoms on the diamond surface transiently. temperature hydrothermal solution which results in the pres- It may be considered that the dangling bond will recombine ence of 575 nm band in Group-B carbonado. The crustal with hydrogen atoms, but it is probable that the dangling inclusion minerals and PAH (polycyclic aromatic hydrocar- bond of carbon on a diamond microcrystal recombines with bon) could have been incorporated into carbonado, and flo- that on the other diamond microcrystal because their grain rencite and some clay minerals as kaolinite ( TRUEB and DE sizes are pretty small and the ratios of surface to volume are WYS, 197 1) would have been formed in this hydrothermal large. This process results in direct bonding between diamond etching event, most likely as the secondary sedimentary microcrystals with the formation of the damaged centers in products. The gentle slope of REE patterns in heavy REEs diamond crystal. seen in A2, AB, Bl, and B2 and the negative europium anomaly suggest the incorporation of crustal materials whose Hydrothermal Etching REE profiles are horizontal in the region of heavy REEs commonly from terbium (or holmium) through lutetium In this section, our attention is directed to one of the most ( HASKIN et al., 1968). Furthermore, the hydrothermal etch- distinctive characteristics of carbonado; the presence of min- ing might have induced changes in carbon isotopic compo- erals of crustal origin ( TRUEB and DE WYS, 197 1). Inclusion sition of carbonado either by incorporation of biological or- minerals characteristic of the upper mantle assemblages ganic carbon carried by the solution or by isotope exchange commonly found in diamonds of kimberlite-origin have not reaction between microdiamonds constituting carbonado and been reported for carbonado under consideration. As men- organic carbon. In either way, the fine crystallite size of dia- tioned above, it is inferred that the carbonado was originally mond in carbonado would have been favorable in enhancing produced through the crystallization of source diamond the isotope exchange reaction. If the hydrothermal etching crystal in the upper mantle followed by binding of these mi- proceeds further, the micropores would be denser and the crocrystals in the crust under the exposure to alpha particles, REE abundance pattern might become similar to that of i.e., without high pressure and high temperature. For further crustal materials. Therefore, it is reasonable to expect that discussion on the origin of carbonado, especially on the pres- other types of carbonado with low REE content might exist. ence of crustal inclusion, one must consider Group-B car- In any event, the genetic origins of forming carbonado would bonado with the 575 nm luminescence which has not been be almost the same among the various types of carbonado, discussed in the previous considerations. Judging from the and the variation of their physicochemical properties such similarity in the appearances between Group-A carbonado as photoluminescence profile and microstructure was deter- and Group-B carbonado, it is likely that these two types of mined by the last event of the history of development, i.e., carbonados had a common forming process; the crystalli- the hydrothermal etching. zation of micro-diamond in the upper mantle and the sec- Genetic History of Carbonado ondary hardening in the crust. The presence of hematite on the surface of carbonado in- On the basis of the various peculiar properties of carbonado dicates that the last stage of the forming environment was mentioned above (observed properties of Central African oxidic, which is consistent with the decrease in U concentra- carbonado are summarized in Fig. 7 ) , one can infer the pos- tion ( OZIMA et al., 199 1) in contrast with the quantity of the sible genetic history of carbonado as follows. In the first place, detected fission product. Uranium ion is highly soluble in microdiamond crystallite as a source crystal of carbonado water as a form of [ UOZ] *+ under the oxidic environment was formed at upper mantle under high pressure and high Radiation damage and REE profiles of carbonado diamonds, Central African Republic 2631

transition from Group-A to Group-B was found to arise between 400 and 500°C. - (RdiationcxposIue) Also the difference in thermal properties between Group- A and Group-B was observed on TG-DTA curves. Group- A carbonado has a narrow exothermic peak at 800°C with a substantial decrease of the weight, while Group-B carbonado gives a broad exothermic band at 790°C with gradual lowering of the weight. Carbonado diamonds give 1384 cm-’ absorption band in IR spectra for the both types of stones, which means the presence of nitrogen platelets in diamond lattice ofcarbonado. It suggests the high temperature origin ofthe seed microcrystal of carbonado probably generated in the upper mantle. No significant difference can be detected in REE isotopic compositions between carbonado and terrestrial standard sample. The absence of detectable effect of uranium fission on REE isotopes is due to the large amount of primordial REEs, whose chondrite-normalised pattern resembles that of kimberlite closely. The much higher REE content in Group- A carbonado than that in Group-B carbonado suggests the difference in the degree of hydrothermal etching between the two groups. On the basis of the observed results, we inferred one of the possible genetic histories of carbonado, that is, crystallization of microdiamond by high pressure and high temperature in the upper mantle, tight binding of the microcrystals in the crust under the influence of radiation from uranium and thorium, and the hydrothermal etching resulting in the variety of physicochemical properties of carbonado. FIG. 7. Block diagram showing the relationship between the chemical properties of carbonado from Central Africa and its inferred Acknowledgments-We are grateful to Professor H. Shimizu of Ku- genesis. mamoto University, Professor M. Ozima of the Osaka University, and Dr. K. Shibata of Geological Survey of Japan for their critical discussions and encouragements. We are greatly indebted to Professor temperature, and nitrogen atoms in the diamond microcrys- Y. Hirose of Tokai University for CVD diamonds grown by the combustion flame method. Our thanks are extended to Dr. M. B. Shabani tals were aggregated to form platelets by annealing. Next, in of University of Tokyo, and Drs. M. Kubota and N. Nonose of Na- the crust the microcrystals were exposed to LYp articles emitted tional Chemical Laboratory for Industry for their technical guidance from uranium and thorium at mild temperature and pressure. and help for REE abundance determination. This paper was greatly This process resulted in the genesis of radiation-damaged improved by helpful reviews from Drs. John Gurney, Henry Meyer, and Chris Koeberl. This study was supported by a Grant-in-Aid for color centers and tight binding of microcrystals of carbonado. Encouragement of Young Scientist, Contract Number 04780060 and Finally, they were etched by hydrothermal fluid, which gave 05854042, from the Ministry of Education, Science and Culture of rise to the elution of uranium ion and portion of the REEs Japan. contained in inclusions, the incorporation of minerals of crustal origin and the variety of the thermal properties and Editoriul handling: S. R. Taylor photoluminescence of carbonado as exhibited by Group-A and Group-B. REFERENCES

CONCLUSIONS AKAGI T. and MASUDAA . ( 1988) Isotopic and elemental evidence for a relationship between kimberlite and Zaire cubic diamonds. All carbonado samples investigated in this study exhibited Nature 336, 665-667. BARRYJ . C. ( 1991) HRTEM of { 100 ] Platelets in natural type 1aA very intense photoluminescence bands resulting from radia- diamond at 1.7 8, resolution: A defect structure refinement. Phil. tion-damaged defects, and the spectral profiles can be clas- Mug. A64, 11 I-135. sified into three groups. Group-A carbonado reveals a pho- BROOKINGSD . G. ( 1984) Geochemical aspects of radioactive waste toluminescence band at 504 nm (3H center), while Group- waste disposal. Springer Verlag. B carbonado reveals a photoluminescence band at 575 nm CHRENKOR . M., TUFT R. E., and STRONGH . M. ( 1977) Trans- formation of the state of nitrogen in diamond. Nufure 270, 14 I- with weak broad bands at 5 12 and 580 nm. The intermediate 144. type of carbonado between Group-A and Group-B was also CLACKSONS . G., MOORE M., WALMSLEYJ . C., and WOODSG . S. found (Group-AB). ( 1990) The relationship between platelet size and the frequency Taking account of the change in photoluminescence spec- of the B’ infrared absorption peak in type Ia diamond. Phil. Mug. 25, 115-128. tra in the heating experiment of Group-A carbonado, it is COLLINSA . T. and LAWSONS . C. ( 1989) Cathodoluminescence proved that Group-B carbonado has been exposed to higher studies of isotope shifts associated with local&d vibrational modes temperature than Group-A carbonado. Experimentally, the in synthetic diamond. J. Phys. Condens. Matter 1, 6929-6937. 2638 H. Kagi et al.

CULLERS R. L. and GRAF J. L. ( 1984) Rare earth elements in igneous LAX M. and BURSTEINE . ( 1955) Infrared lattice absorption in ionic rocks of the continental crust: Predominantly basic and ultrabasic and homopolar crystals. Phys. Rev. 97, 39-52. rocks. In Rare Earth Element Geochemistry(ed. P. HENDERSON), MITCHELLR . H. ( 1986) Kimberlites: Mineralogy, Geochemistry, and pp. 237-214. Elsevier. Peterology. Plenum. DAVIESG . ( 1977) The optical properties of diamond. Chem. Phys. MORIYOSHIY ., KAMO M., SETAKAN ., and SATO Y. (1983) The Carbon 13, 1-144. microstructure of natural polycrystal diamond, carbonado and DEANP . J. ( 1965) Bound excitons and donor-acceptor pairs in natural ballas. J. Mater. Sci. 17, 217-224. and synthetic diamond. Phys. Rev. 139, A588-602. OZIMAM ., ZASHUS ., TOMURAK ., and MATSUHISAY . ( 199 1) Con- DYER H. B.. RAALF . A.. DU PREEZL .. and LOUBSERJ . H. N. ( 1965) straints from noble-gas contents on the origin of carbonado dia- Optical absorption features associated with paramagnetic nitrogen monds. Nature 351,472-474. in diamond. Phil. Mag. 11, 763-764. PATE B. B. ( 1986) The diamond surface: Atomic and electronic EVANST . and PHAALC . ( 1962) Imperfection in type I and type II structure. Surface Sci. 165, 83-142. diamonds. Proc. Roy. Sot. London A210,538-552. ROBINSOND . N. ( 1978) The characteristics of natural diamond and EVANST . and QI Z. ( 1982) The kinetics ofthe aggregation of nitrogen their interpretation. Mineral. Sci. Eng. 10, 55-72. atoms in diamond. Proc. Roy. Sot. London A381, 159-178. SMIRNOVG . I., MOFOLO M. M., LEROTHOLIP . M., KAMINSKY FRANTSESSONE. V. and KAMINSKII F. V. ( 1974) Carbonado, a dia- F. V., GALIMOVE . M., and IVANOVSKAY1A. N. ( 1979) Isotopically mond variety of nonkimberlitic origin. Dokl. Akad. Nauk SSSR light carbon in diamonds from some kimberlite pipes in Lesotho. 219, 187-189 (in Russian). Nature 278,630. GALIMOVE . M., KAMINSKII F. V., and KODINAL . A. ( 1985) New SMITH V. J. and DAWSONJ . B. (1985) Carbonados; Diamond ag- data on isotope composition of carbonado. Geokhimia 5, 723- gregates from early impacts of crustal rocks? Geology 13, 342- 726 (in Russian). 343. HALLH . T. ( 1970) Sintered diamond: A synthetic carbonado. Science SOBOLEVE . V., LISOIVANV . I., and LENSKAYAS . V. ( 1968) X-ray 169, 868. diffraction spikes as related to optical properties of natural dia- HAMZAA . V., KUBIAK G. D., and STULENR . H. (1988) The role monds. Dokl. Acad. Nauk. SSSR. 175, 665-668 (in Russian). of hydrogen on the diamond C( 1 1 1 )-( 2 X 1) reconstruction. Sur- SOBOLEVN . V. and SHATSKYV . S. ( 1990) Diamond inclusions in face Sci. 206, 833-844. garnets from metamorphic rocks: A new environment for diamond HASKIN L. A., HASKIN M. A., FREY F. A., and WILDEMANT . R. formation. Nature 343, 742-746. (1968) Relative and absolute terrestrial abundances of the rare SUTHERLANDG . B. B. M., BLACKWELLD . E., and SIMERALW . G. earths. In Origin and distribution of‘the elements (ed. L. H. AH- (1954) The problem of the two types of diamond. Nature 174, RENS). pp. 889-912. Pergamon. 901-904. HIDAKA H. and MASUDAA . ( 1988) Nuclide analysis of rare earth SWINDLED . L., WRIGHT R. J., and KURODA P. K. ( 1971) Yields element of the Oklo uranium ore samples: a new method to esti- of ruthenium isotopes from the spontaneous fission of *‘*U. J. mate the neutron fluence. Earth Planet. Sci. Lett. 88, 330-336. Inorg. Nucl. Chem. 33, 876-879. HIRATAT ., SHIMIZUH ., AKAGI T., SAWATARHI ., and MASUDAA . TRUEB L. F. and BUTTERMANW . C. ( 1969) Carbonado: A micro- ( 1988) Precise determination of rare earth elements in geological structural study. Amer. Mineral. 54, 4 12-425. standard by inductively coupled plasma source mass spectrometry. TRUEB L. F. and DE WYS E. C. (1969) Carbonado: Natural poly- Anal. Sci. 4, 631-643. crystalline diamond. Science 165, 799-802. HIROSEY ., AMANUMAS ., and KOMAKIK . ( 1990) The synthesis of TRUEB L. F. and DE WYS E. C. ( 1971) Carbon from Ubangi: a high-quality diamond in combustion flames. J. Appl. Phys. 68, microstructural study. Amer. Mineral. 56, 1252-1256. 640 I-6404. VIDALIG . and FRANKLD . R. ( 1983) He-diamond interaction probed IWAKIM ., TAKAHASHIK ., and SEKIGUCHAI . ( 1990) Wear property by atom beam scattering. Phys. Rev. B 27, 2480-2487. and structure of nitrogen implanted glassy carbon. J. Mater. Rex VIDALIG ., COLE M. W., and WEINBERGW . H. (1983) Helium as 5,2562-2566. a probe of the ( I I 1) surface of diamond. Phvs. Rev. Left. 51, 118- KAGI H., MASUDAA ., and TAKAHASHIK . (1991) Laser-induced 121. luminescence from natural polycrystal diamond, Carbonado. Nu- VINOCRADOVA . P., KROPOTOVA0 . I., ORLOV Yu. L., and GRI- turwissenschajien 18, 355-358. NENKO V. A. ( 1966) Isotopic compositions of diamond and car- KAMINSKII F. V. ( 1987) Genesis of carbonado; polycrystalline ag- bonado crystals. Geochem. Intl. 3, 11 23- 1125. gregate of diamond. Dokl. Akad. Nauk SSSR 291, 439-440 (in WACLAWSKBI . J., PIERCED . T., SWANSONN ., and CELOTTAR . J. Russian ). ( 1982) Direct verification of hydrogen termination of the semi- KAMINSKII F. V., KULAKOVAI . I., and OGLOBLINAA . I. (1985) conducting diamond ( 1 I1 ) surface. J. Vat. Sci. Technol. 21,368- About polycyclic aromatic hydrocarbons in carbonado and dia- 370. mond. Dokl. Akad. Nauk SSSR 240,985-988 (in Russian). WALKERJ . ( 1975) Optical and uniaxial stress studied of irradiation- KANDA H., SUZUKI K., FUKUNAGAO ., and SETAKA N. (1976) induced defects in diamond. In Proceedings of the International Growth of polycrystalline diamond. J. Mater. Sci. Left. 11,2336- Coqltirence on Lattice D&cts in Semiconductors (ed. F. A. 2338. HUNTLEY) , 3 17. Inst. Phys. KIRKLEY M. B., GURNEYJ . J., OTTER M. L., HILL S. J., and DANIELS L. R. ( 199 1) The application of C isotope measurements to the WALKERJ . ( 1979) Optical absorption and luminescence in diamond. identification of the sources of C in diamonds: A review. Appl. Rep. Prog. Phys. 42, 1605-1659. Geochem. 6,477-494. WENTORFR . N. and BOVENKERK( 196 I ) On the origin of natural KLYUEV Yu. A., NEPSHA V. I., EPISHINAN . I., SMIRNOVV . I., diamonds. Astrophys. J. 134, 995-1005. PLOTNIKOVAS . P., PROKOPCHUKB . I., and KAMINSKIIF . V. WILKSJ . and WILKS E. M. ( 1992) Wear and polishing of diamond. ( 1978) Structural peculiarities of natural polycrystalline diamonds. In Properties q/natural and synthetic diamond (ed. J. E. FIELD), Dokl. Akad. Nauk SSSR 240, 1104- 1107 (in Russian). pp. 574-602. Academic.