Volume expansion of diamond during ion implantation at low temperatures

Volume expansion of diamond during ion implantation at low temperatures

Nuclear Instruments and Methods in Physics Research Blg (1987) 261-263 North-Holland, Amsterdam VOLUME EXPANSION OF DIAMOND AT LOW TEMPERATURES J.F...

250KB Sizes 0 Downloads 29 Views

Nuclear Instruments and Methods in Physics Research Blg (1987) 261-263 North-Holland, Amsterdam

VOLUME EXPANSION OF DIAMOND AT LOW TEMPERATURES

J.F. P R I N S *, T . E . D E R R Y

DURING

261

ION IMPLANTATION

a n d J.P.F. S E L L S C H O P

Wits-CSIR Schonland Research Centre/or Nuclear Sciences. University of the Witwatersrand. Johamlesburg. 2000, South Africa

Received 9 September 1986

A type IIa diamond was implanted with fluorine ions while being maintained at liquid nitrogen temperature (77 K). After allowing the diamond to warm to room temperature the volume expansion caused by the ion implantation was measured using standard profilometry. No expansion could be detected up to an ion dose 6 • l0 ts ions/cm 2. Above this ion dose, expansion, which could be ascribed to amorphization of the ion damaged layer towards a graphite phase, did occur. This result is in severe contrast to ion implantation carried out above room temperature where expansion occurs immediately with the onset of ion implantation, and can be explained in terms of the diffusion of interstitials out of the ion damaged layer leaving behind a high density of immobile vacancies.

1. Introduction Ion implantation of diamond with carbon ions while keeping the diamond at a suitably elevated temperature leads to internal diamond growth [1,2]. At lower temperatures, implantation to high ion doses causes the diamond to blacken owing to the increase in radiation damage [3,4]. This blackening may be a result of amorphization setting in, in the highly damaged implanted layer. It has also been established that ion implantation under the latter conditions causes volume expansion of the implanted diamond [5,6]. At high enough ion doses, this expansion manifests itself as a surface protrusion which is measurable by standard profilometry. F r o m their study, Maby, Magee and Morewood [5] suggested that the volume expansion may be caused by the onset of amorphization which induces the metastable diamond to move towards graphitization. The expansion is then attributable to the difference in densities between diamond and graphite. This process would limit the extent to which diamond could be effectively doped by ion implantation. Our subsequent study on ion implantation of fluorine ions into diamond [6] indicated that expansion occurs immediately with the onset of ion implantation without first having to reach a certain critical ion dose. The conclusion was reached that the expansion is not caused by amorphization setting in nor by a tendency for the implanted layer to revert to graphite, and that the explanation should be * On secondment from De Beers Diamond Research Laboratory, P.O. Box 916, Johannesburg, 2000, South Africa. 0168-583X/87/$03.50 9 Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)

sought in terms of the creation and interaction of the standard radiation defect products obtained during ion implantation. Based on the results found during another study on the hopping conduction of ion implanted diamond [7], the volume expansion could be modelled adequately by taking into account the subsequent interaction of interstitial atoms and vacancies after their creation in the collision cascades. According to this model [6], the observed expansion occurs when the target temperature allows interstitial atoms, or at least some of them, to diffuse, but is low enough to leave the vacancies immobilized. Those diffusing interstitials which egress from the ion damaged layer before recombining with a vacancy, can then move to sinks, such as the nearby diamond surface, and in this way add to the diamond volume. The remaining implanted layer is thus essentially a vacancy rich crystal lattice which to a large extent retains its gross integrity as diamond. If this model is correct, implantation at a temperature which also limits interstitial atom diffusion, should severely inhibit the expansion observed. In the literature, considerable disagreement is found about the temperature at which the self-interstitial in diamond becomes mobile. From ESR measurements at low temperature on electron irradiation of diamond, Lomer and Marriott [8] suggested that the self-interstitial may become mobile at temperatures as low as 50 K. In a subsequent study Flint and Lomer [9] concluded that the annealing stage observed at 50 K can also be explained by the movement of light impurities, such as hydrogen, during electron irradiation and not necessarily by interstitial movement at this low temperature.

262

J.F. Prins et aL / Volume expansion of diamond

Prdviously Bourgoin, Massarani and Visocekas [10] had observed the annealing stage at 50 K, but concluded that it was a charge transfer process and not caused by defect annealing. The first recovery stage which can be ascribed to interstitial atom diffusion was observed at 260 K by Massarani and Bourgoin [11]. Measurement of the activation energy for the recovery of this defect gave 0.13 eV. This is the same as the energy measured by Clark and Palmer [12] for partial annealing of the GR I centre observed in diamond around 300 and 500~ and believed to be associated with mobile interstitials recombining with vacancies. It is highly unlikely that this low activation energy will allow large scale movement of interstitial atoms at liquid nitrogen temperatures (77 K). The electrical "resistance measured at room temperature for carbon ion implanted diamond layers obtained at a target temperature of 77 K was found to be 106 times higher than that for similar layers implanted at 240~ [13]. This higher resistance may be attributable to interstitial atoms remaining in the implanted layer, and partially compensating the hopping conduction centres. Annealing towards lower resistances occurred in the temperature range of about I00 to 600~ which overlaps the annealing stages observed for the GR I Centre [12] thought to be caused by interstitial movement. Thus, if the volume expansion of diamond is caused by interstitial atom diffusion [6], and if the interstitial atoms only become significantly mobile above room temperature, ion implantation of diamond at liquid nitrogen temperatures should severely curtail the expansion measured afterwards.

2. Experiment and results Use was made of a type IIa diamond on to which a flat area of approximately 150 mm2 was polished. Different spots for irradiation were masked off using a stainless steel disc with a 2 mm diameter hole in it, and each spot implanted at liquid nitrogen temperature (77 K) using 170 keV fluorine ions at a dose rate of 2.2 • 1013 ions cm -2 s -l. After each implantation, the diamond was allowed to warm up to room temperature, when the height of the implanted spot was measured with a Taylor-Hobson Surtronic 3 profilometer. The results obtained are displayed in fig. 1 and compared with the results which were measured previously after implantation at ambient temperature [6]. Whereas at ambient temperature the expansion was immediate and most rapid at low ion doses, the low temperature implantating does not cause any detectable expansion up to an ion dose of 6 x 10 j5 ions/cm-'. Between an ion dose of 6 • t5 and 1 • ~6 ions "l cm -, a rapid increase in volume initiates which seems

0.18 0.16 .

~~, 0.12 0.10 0.08

~- 0.06u~

o ROOM TEMP

O.Ot,

0.02

-

-

0 9 10N DOSE ( x 1016ions - crn-Z) Fig. 1. The volume expansion obtained when implanting diamond with fluorine ions while being held at 77 K, compared to the volume expansion obtained when implanting under the same conditions at room temperature.

to level off at a value of approximately 0.07 gm, above 2 • 1016 cm-2. Above 5 x 10 t6 cm 2 the step height increases again more rapidly to saturate finally at a value of about 0.13 gin.

3. Discussion Ion implantation at liquid nitrogen temperature evidently changes the whole mechanism causing volume expansion during ion implantation. In this case expansion occurs only after a certain critical stage of radiation damage has been achieved. The absence of any detectable expansion at doses below this ion dose supports the theory that the immediate expansion observed at low ion doses when implantating at temperatures above ambient, is caused by interstitial atoms diffusing out of the ion damaged region leaving behind a vacancy rich layer of lower density [6,7]. In the low temperature case, it seems reasonable to conclude that the onset of expansion above the critical ion dose may be caused by amorphization of the ion damaged layer at high ion doses, as originally proposed by Maby, Magee and Morewood for their room temperature results [5]. Owing to its metastability, the amorphized diamond will then revert to a graphitic form. With increasing ion dose, the whole width of the ion damaged layer should eventually reach this graphitic state. Using the program TRIM-86 [14], the distribution of total target vacancies expected when implanting diamond with 170 keV fluorine ions could be obtained, and the total width determined as 0.21 p.m. If this whole

J.F. Prins et al. / Vohtme e_xpansion of diamoml

layer converts to graphite, i.e. changes its density from 3.51 g/cm 3 to 2.24 g / c m 3, the step height caused by volume expansion should be 0.12 #m, which corresponds closely to the 0.13 #m saturated step height measured. (See fig. 1.) The shape of the expansion curve could be caused by the shape of the point defect distribution obtained when implanting diamond. According to TRIM-86 [14] the point defect distribution consists basically of two superposed parts: a Gaussian distribution centered around a depth of 0.14/~m having a fwhm of approximately 0.10 # m and a long tail containing a lower density of defects which extends out to the surface, causing the total damage width to be the 0.21 #m already mentioned. If a certain critical density of defects is needed to initiate expansion, this will occur sooner under the Gaussian peak than in the lower density tail. Using the fwhm width of 0.10 and assuming that this width of damaged material reverts to graphite before the tail does, leads to a step height of 0.06/~m. This is close to the value of 0.07 #m where the step height increase first levels off. The temperature at which the diamond is held during ion implantation should have a large effect on the subsequent annealing behaviour of the radiation damage introduced. When implanting at a temperature above ambient, causing that type of volume expansion, the remaining damaged layer will contain a large number of vacant atomic sites. The number of vacancies may even exceed 30% [6]. Annealing to high temperatures will then surely favour a transition to graphite. After implantation at liquid nitrogen temperature (below the ion dose at which expansion occurs) the damaged layer will still contain the vacancies and interstitials produced in the collision cascades. A high density of these defects will favour vacancy-interstitial recombination when annealing the diamond at high temperature where point defect diffusion can occur. Accordingly fewer interstirials will escape leaving a high density region favouring the diamond phase. The latter possibility was exploited by Prins and Raal [15] to change the optical absorption spectrum of a type IIa diamond to a spectrum containing type Ib absorption characteristics after carbon and nitrogen ion implantation at liquid nitrogen temperature followed by annealing up to 800 o C. This change in absorption spectrum indicated that the implanted layer contained substitutional nitrogen atoms in the diamond lattice.

3. Conclusion Below a certain critical ion dose, implantation of diamond while held at liquid nitrogen temperature does

263

not cause the large volume expansion observed when implanting to the same doses above room temperature. This lends further support to the theory that the latter volume expansion is caused by diffusion of interstitials out of the ion-damaged layer, in this way leaving behind an excess of immobile vacancies [6,7]. This investigation is part of a programme financed by De Beers Industrial Diamond Division at the Schonland Research Centre for Nuclear Sciences. It was initiated by Dr H.B. Dyer and Dr C. Phaal of De Beers in cooperation with the University of the Witwatersrand. We are also indebted to Mr M. Rebak who provided technical assistance in preparing the diamond and measuring the protrusions. Dr R.W. Fearick and Mr P.E. Harris obtained data from the TRIM-86 program and Mr A. Mashabela operated the ion implanter.

References [1] R.S. Nelson, J.A. Hudson, D.J. Mazey and R.C. Piller. Proc. R. Soc. London, Ser. A. 386 (1983) 211. [2] T.E. Derry and J.P.F. Sellschop, Nucl. Instr. and Meth. 191 (1981) 23. [3] V.S. Vavilov, V.V. Krasnopevtsev and Yu.V. Milyutin, Radiat. Eft. 22 (1974) 141. [4] R. Kalish, T. Bernstein, B. Shapiro and A. Talmi, Radiat. Eft. 52 (1980) 153. [5] E.W. Maby, C.W. Magee and J.H. Morewood, Appl. Phys. Left. 39 (1981) 157. [6] J.F. Prins, T.E. Derry and J.P.F. Sellschop, to be published in Phys. Rev. B. [7] J.F. Prins, Phys. Rev. B31 (1985) 2472. [8] J.N. Lomer and D. Marriot, Inst. Phys. Conf. Ser. 46 (1979) 341. [9] I.T. Flint and J.N. Lomer, Physica l16B (1983) 183. [10] J. Bourgoin, B. Massarani and R. Visocekas, Phys. Rev. B18 (1978) 786. [11] B. Massarani and J.C. Bourgoin, Phys. Rev. B14 (1976) 3682. [12] C.D. Clark and D.W. Palmer, Proc. 4th Symp. Reactivity of Solids (Elsevier, Amsterdam, 1960) p. 436. [13] J.F. Prins, Radiat. Eft. Lett. 76 (1983) 79. [14] J.F. Ziegler, J.P. Biersack and U. Littmark, The Stopping and Range of Ions in Solids, (Pergamon, New York, 1985). [15] J.F. Prins and F.A. Raal, to be published in Radiat. Eft.