Pii: s0168-583x(01)01022-9

Nuclear Instruments and Methods in Physics Research B 188 (2002) 78–83 Study of the irradiation damage in SiC by ion channeling M. Kokkoris a,*, S. Kossionides a, A. Kyriakis a, K. Zachariadou a, G. Fanourakis a, R. Vlastou b, Th. Paradellis a a Laboratory for Material Analysis, Institute of Nuclear Physics, NCSR ‘Demokritos’, GR-153 10 Aghia Paraskevi, Athens, Greece b Department of Physics, National Technical University of Athens, GR-157 80, Athens, Greece The importance of silicon carbide as a wide band gap semiconductor is widely accepted and well documented. Its excellent physical properties (chemical inertness, high-temperature strength, low thermal expansion, extreme hardness)make it the most promising substitute for traditional semiconductors, especially when high-temperature, high-voltagepower, and high-frequency devices are concerned.
In the present work, the gradual amorphization of a SiC Lely (21R) crystal when irradiated with 8 MeV 7Li ions in a random direction up to a maximum dose of approximately 1 Â 1016 particles/cm2, is being studied, using the progressivechange of channeling parameters for different depths. The results refer to the energy region of $1 MeV/nucleon, and anattempt is made in order to explain the peculiarities of the experimental spectra and the mechanism of defect productionin SiC. As in previous studies, a change in color was observed after irradiation in the random mode, indicating that theproblem of irradiation damage in SiC caused by light ion beams needs further investigation. Ó 2002 Elsevier ScienceB.V. All rights reserved.
PACS: 61.80.-xKeywords: Backscattering; Channeling; SiC crystal; Lithium; Lely; RBS semiconductors. There exists a growing interestin the physical nature of the effect of this high- The effects of irradiation on the properties of energy ion implantation as it is considered to be a solids are of significant interest in scientific and promising way of increasing the microelectronic technological context. Several studies have been chip integration by the formation of multilayer presented concerning irradiations with ions having three-dimensional structures [4]. Moreover, there an energy of the order of 1 MeV/nucleon [1–3].
are strong indications that the increase of ion These ions find an ever increasing application in energy does not lead exclusively to quantitative the modification of the properties of metals and changes of the ion implanted layer parameters butalso to qualitative changes of the defect-impuritystructure of the whole irradiated area [5].
This type of implantation is mainly character- * Corresponding author. Tel.: +30-1-4288217; fax: +30-1- ized by the high linear density of the energy con- tribution by ions into the electronic subsystem of E-mail address: [email protected] (M. Kokk- the target, resulting to electronic energy losses of 0168-583X/02/$ - see front matter Ó 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 8 - 5 8 3 X ( 0 1 ) 0 1 0 2 2 - 9 M. Kokkoris et al. / Nucl. Instr. and Meth. in Phys. Res. B 188 (2002) 78–83 the order of MeV/lm. It is also characterized bythe fact that this type of losses strongly prevailover the direct transmission of the ion energy tothe nuclear subsystem of the target, which is pre-dominant at the end of the mean ion projectedrange but almost negligible otherwise (<0.5% inour case as shown with the use of the TRIM code).
It should be noted here however, that the data on defect production and annealing at depths ofthe order of a few lm inside a target are sparse,often contradictory and by no means conclusive asto the very nature of the mechanism of the phe-nomenon. On the other hand, there is a strongongoing effort, evident in the literature, for theinvestigation of different SiC characteristics, such Fig. 1. RBS/C spectra at Ep ¼ 1:5 MeV, showing the excellent as structural and electrical properties [6–9], radia- crystalline quality of the Lely 21R SiC target (v tion damage [10–15] and the effects of aluminum, integrated over Æ2 channels after the surface peak).
boron, oxygen and nitrogen implantation for p-type and n-type doping, as well as for surface defect density ($1:5–3 Â 1018 cmÀ3). Protons at 1.5 passivation and electrical isolation [6,16].
MeVwere used in order to align the beam to the This work is an attempt to present the damage (0 0 0 1) axis of the crystal. The fine-tuning of the induced by high dose irradiation of 8 MeV 7Li2þ channeling position was finally achieved via an- ions, in the case of a SiC Lely platelet (21R) ex- gular scans, which revealed the excellent crystalline hibiting excellent crystalline behavior, and analyze it by means of the progressive change of channel- order of 2–4% (in an integrated region of Æ2 chan- nels after the surface peak), as shown in the RBS/Cspectra at Ep ¼ 1:5 MeV, presented in Fig. 1.
After the crystal alignment, the experiment (a) The experiment is carried out through irradia- ‘‘Demokritos’’, Athens, Greece, using the 5.5 MV tions with the 8 MeV 7Li2þ ions in the random TN11 TANDEM accelerator. The final ion energy direction (which is achieved via a random rota- was determined via NMR with an estimated error tion of the sample during the spectrum acqui- of 5 keV. Lithium 2þ ions were accelerated to an energy E ¼ 8 MeVand were lead to a scattering (b) At the very beginning (for normalization pur- chamber, which included a 4-motor goniometer poses), as well as after each dose step in the system capable of determining the target orienta- tion with an accuracy of 0.01°. The detection sys- ($2 Â 1014 particles/cm2) in the channeling di- tem consisted of a single Si surface barrier detector rection using the same beam, in order to check having an overall resolution of 12 keVfor a-par- the progressive change in the v values (with v ticles (8 keVfor protons). The beam divergence was less than 0.07° due to the long collimation beam, or C/RBS per channel ratio) at different system. The beam spot size was %2 Â 3 mm2 and the beam current did not exceed 5 nA on target.
(c) The results are analyzed with the RUMP code, The SiC target used was a Lely platelet (by Si- the spectra are compared and the correspond- Crystal AG), cut in the (0 0 0 1) orientation, with a ing normalized v values are extracted (as ana- 21R polytype structure, having a relatively low lyzed in the following section). Subsequently, M. Kokkoris et al. / Nucl. Instr. and Meth. in Phys. Res. B 188 (2002) 78–83 the curves v ðxÞ ¼ f (dose) are studied at vari- ous depths x for the crystal into consideration.
(d) After the irradiation with the lithium ions, a test is performed with protons at Ep ¼ 1:5MeV. Channeling spectra from the irradiatedspot and from a virgin area are comparedand the v values are extracted. The advantageof the proton beam is that it can probe up to adepth of %10 lm, close to the end-of-range ofthe lithium ions (12.3 lm) and thus examinethe damage inflicted on the crystal as well asits possible amorphization for the total incor-porated dose.
During the course of data acquisition, severalsources of error affect the precision of the mea- Fig. 2. v as a function of dose for depths of 0.1 and 0.5 lm surements. The exact dimensions of the beam spot inside the target. The error bars (15%) are indicated in the have an utmost importance in the accumulated graphs. The results correspond to the Si-phase.
dose per step and any small deviations can lead touncertainties. During the irradiation one cannotexclude a slight change in the position of the beamspot, resulting in small oscillations of the calcu-lated v values. Moreover, the resolution of thesilicon surface barrier detector in medium massions, such as Li, is mediocre ($40 keV).
The results from the irradiated SiC crystal are presented in Figs. 2 and 3, where the progressivechanges of the v values at different depths inside each target (0.1, 0.5 and 3.3 lm respectively), rel-ative to the accumulated dose are recorded; v is Fig. 3. v as a function of dose for a depth of 3.3 lm inside the target. The error bars (15%) are indicated in the graph. Theresults correspond to the Si/C-phases.
where vð0Þ corresponds to the virgin crystal [17].
The fitted curves presented are empirical (of exponential type), since there is no generally ac- power) in both the random and the channeling cepted multiparameter formalism concerning the direction, which introduces an uncertainty in the phenomenon. The total estimated errors are also depth determination of the order of 5–10%. Data in the bibliography concerning such parameters It should be noted here that there exist three for channeled ions in semiconductors and metals major drawbacks in the trend of analyzing the are sparse and concern experiments that have been carried out in the transmission and not in the The first drawback is connected to the as- backscattering geometry. To the authors’ best sumption of the same Se (electronic stopping knowledge, especially for SiC, very few relative M. Kokkoris et al. / Nucl. Instr. and Meth. in Phys. Res. B 188 (2002) 78–83 works have been presented [8]. If one uses a light accumulated dose, the damage induced in Si was ion beam (e.g. protons) in order to examine the greater (by $25%) than the one induced in C, in progressive change of the v values, this uncertainty accordance to experimental data (vacancies pro- in the depth determination is enhanced. On the other hand, the use of the lithium beam has the The parameter v for a damaged crystal con- disadvantage of poorer statistics and the danger to tains components corresponding to imperfections cause a small but not negligible extra damage of in the virgin crystal, plus the yield of scattering on statically displaced atoms, plus the change due to The second drawback is related to the second- dechanneling during a particle’s path through the ary electron suppression, which was achieved with damaged crystal layer. Following the reasoning the application of a positive voltage directly on the developed in [17], and the simplest model pre- target, along with a permanent magnetic field of dicting saturation of defect concentration at a level 0.01 T. This suppression is far from being perfect.
below 1, known from [21], the following equations Ions of medium mass, such as lithium, hitting on a target can cause the escape of a large number ofsecondary electrons in the backscattering direc- tion, thus requiring a large correction factor for ¼ ½1 À vð0ފ½1 À nDðdoseފ½1 À F ðnDsފ; the normalization of the charge. There is substan-tial evidence [18,19] suggesting that the second- nDðdoseÞ ¼ ½R=ðR þ b þ aފ½1 À eÀðRþbþaÞtŠ; ary electron emission in the channeling directionis strongly related to the dE/dx of the incoming ion.
where F is the dechanneling component being de- Thus, the combined statistical and systematic pendent on nD(dose) and the pathlength s, R is the error cannot be less than 10–15% in the most fa- rate of displacement due to atomic collisions, a(T ) vorable case, therefore only effects significantly is the coefficient of the temperature induced re- exceeding the above mentioned value can be reli- combination of free defects, and b is a coefficient ably identified. Such effects have not been ob- which accounts for additional annealing due to the electronic energy losses, Se, of medium mass ions.
The third drawback is connected to the sepa- This coefficient bðSeÞ, being dependent on the ion ration of the Si/C phase. Close to the surface the species, can describe nD(dose) functions for all inflicted damage can be attributed exclusively to ions. If, to a first-order approximation, the term F the Si phase from the RBS/C spectra, while at is considered to be constant, then it is evident from greater depths the two phases cannot be separated.
Eqs. (1) and (2) that nD(dose) is roughly propor- It could be deduced though [20] that the observed tional to the parameter v . Thus, following Eq. (3), changes in the v values correspond to a greater the response of v with respect to the dose is ex- extent to the Si-phase, since the C-phase signals pected to be exponential, reaching a ‘‘plateau’’, come from smaller depths where the damage in the after which no significant damage with the inflicted crystal structure is considerably reduced.
In an attempt to comprehend the behavior of The results are presented in Figs. 2 and 3. Near the irradiated crystal, simulations were performed the crystal’s surface (Fig. 2, at a depth of 0.1 lm), using the TRIM code, which calculates the in- and for a dose below 6 Â 1015 atoms/cm2 the in- duced damage according to the Kinchin–Pease flicted damage is practically negligible. For a depth model. The nuclear energy loss, especially for the of 0.5 lm (Fig. 2) the damage is small but not depths into consideration, is just a small percent- negligible, while at greater depths it becomes quite age of the electronic one. However, due to the high significant (Fig. 3). The fitting curves show an accumulated dose, a lot of displacements per tar- excellent monotonic behavior in all cases.
get atom (dpa) are produced. Depending on the This response is confirmed with the use of the initial settings of the simulation (displacement and proton beam at Ep ¼ 1:5 MeV(Fig. 4), where the binding energies), it was evident that, for the same unnormalized v values from the irradiated and a M. Kokkoris et al. / Nucl. Instr. and Meth. in Phys. Res. B 188 (2002) 78–83 [10]. This change cannot be attributed to carbonbuildup formation on the target’s surface, as it hasbeen tested microscopically after the irradiation.
The damage induced in a SiC crystalline target by 8 MeV 7Li2þ ions has been studied and ana-lyzed using the progressive change of the chan-neling parameter v. The possible random andsystematic errors have been reported and theirimpact on the quality of the measurements hasbeen discussed. A mechanism was suggested inorder to explain the experimental results.
Nevertheless, a lot of problems seem to be open Fig. 4. Plain v values with respect to depth in the case of 1.5 for discussion and further analysis, namely, the MeVprotons after the irradiation. The two curves correspondto a non-irradiated (virgin) area and to the irradiated spot.
radiation hardness of different semiconductors andthe existence of a precise general formalism de-scribing the phenomena.
virgin spot with respect to depth inside the crystal It is doubtful whether the simple phenomeno- are presented. With the proton beam, the dose- logical model adopted in the present work is valid inflicted damage can be studied up to a depth of for all SiC polytypes, because the microscopic %10 lm. It is clear from the fitting curves that the behavior of point defects in a SiC crystal at finite irradiated sample is expected to be completely temperature under the beam is not considered in amorphized at the end-of-range of the lithium this approach. It is also the authors’ firm belief, ions (12.3 lm). This latter result is confirmed by that a thorough study of the dechanneling and EPR measurements currently under study. The stopping powers of light beams in a variety of SiC non-monotonic dependence of v with respect to polytypes is imperative before a complete under- the accumulated dose, which has been reported for standing of the damage mechanism is accom- Si irradiated with 16 MeV 14N at comparable do- ses [3], has not been observed in the present workfor SiC within experimental errors.
The saturation of defect creation, as seen in Figs. 2 and 3, can be explained mainly by defect [1] M. Toulemonde, E. Balanzat, S. Bouffard, J.C. Jousset, annealing due to high electronic energy loss [17]. A Nucl. Instr. and Meth. B 39 (1989) 1.
second-order effect is the subsequent passage of an [2] R.G. Elliman, J.S. Williams, W.L. Brown, A. Leiberich, ion along the path of another, which can lead to D.M. Maher, R.V. Knoell, Nucl. Instr. and Meth. B 19/20 the breaking of stable defects, and therefore to their annealing. The amount of energy released [3] T.A. Belykh, A.L. Gorodishchensky, L.A. Kazak, V.E.
Semyannikov, A.R. Urmanov, Nucl. Instr. and Meth. B 51 during ion penetration (up to $800 keV/lm) is enough to produce heating of the material near the [4] P.F. Byrne, V.W. Cheung, Thin Solid Films 95 (1982) ion trajectory (thermal spike induced crystalliza- tion and partial restoration of the lattice [22]).
[5] V.S. Varichenko, A.M. Zaitsev, A.A. Melnikov, W.R.
As a possible explanation for the change of Fahrner, N.M. Kasytchits, N.M. Penina, D.P. Erchak,Nucl. Instr. and Meth. B 94 (1994) 259.
color in the irradiated area, the formation of sta- [6] L. Wang, J. Huang, X. Duo, Z. Song, C. Lin, C.-M.
ble (at room temperature) charged point defects in the end-of-range has been proposed in the past M. Kokkoris et al. / Nucl. Instr. and Meth. in Phys. Res. B 188 (2002) 78–83 [7] N. Schulze, D. Barrett, M. Weidner, G. Pensl, Mater. Sci.
[14] E. Morvan, N. Mestres, F.J. Campos, J. Pascual, A.
een, M.K. Linnarsson, A.Yu. Kuznetsov, Mater. Sci.
[15] M. Kokkoris, S. Kossionides, R. Vlastou, X.A. Aslanog- [9] D.M. Brown, E. Downey, M. Grezzo, J. Kretchmer, otzschel, B. Nsouli, A. Kuznetsov, S. Petrovic, V. Krishnamurthy, W. Hennessy, G. Michon, Solid-State Th. Paradellis, Nucl. Instr. and Meth. B 184 (2001) 319.
[16] W. Wesch, Nucl. Instr. and Meth. B 116 (1996) 305.
[10] W. Fukarek, R.A. Yankov, W. Anwand, V. Heera, Nucl.
[17] H. Huber et al., Nucl. Instr. and Meth. B 146 (1998) 309.
[18] H. Kudo, K. Shima, K. Masuda, S. Seki, Phys. Rev. B 43 M.K. Linnarsson, E. Morvan, B.G. Svensson, Mater. Sci.
[19] H. Kudo, K. Shima, K. Masuda, S. Seki, Phys. Rev. B 43 B.G. Svensson, Mater. Sci. Forum 338–342 (2000) Spring Meeting, Strasbourg, France, 2001.
[21] F.L. Vook, H.J. Stein, Radiat. Eff. 2 (1969) 23.
[13] W. Jiang, W.J. Weber, Mater. Sci. Forum 338–342 (2000) [22] S. Furuno, H. Otsu, K. Hojou, K. Izui, Nucl. Instr. and

Source: http://www.inp.demokritos.gr/experimental/iba/pubs/articles/NIMB_188_78_2002.pdf