Results. Some typical fluorescence spectra are shown in fig.1. The EXAFS signal has been extracted using a technique already described elsewhere [5].

Transcription

1 Local structure investigation of amorphous germanium hydrogenated thin films doped with column III and V elements. G.Dalba, P.Fornasini, R.Grisenti, F.Rocca, I.Chambouleyron. Introduction. One of the major debates on the amorphous semiconductor physics relies on the understanding of microscopic mechanisms of substitutional doping in hydrogenated amorphous silicon and germanium thin films doped with group III and V elements. The hydrogenated amorphous germanium (a-ge:h) is particularly interesting, due to its narrow gap, for the production of infrared sensitive devices. Most of the potential advantages, however, rely on the ability of controlling the electronic properties of the active semi-conducting layer by chemical doping. One of the most important aspects of this process is concerned with the chemical role of the dopant species. This is an important question since most physical properties of the amorphous semiconductors are determined by the local energy minimisation involving network relaxation around individual atoms. So, it is expected that different impurities will exhibit different distribution of co-ordination states. A further aspect to be clarified is how the fraction of impurities, as well as the doping-induced effects, are related to the doping efficiency. In this work, amorphous germanium thin films with different doping elements and concentrations have been investigated by EXAFS spectroscopy. These preliminary results constitute a first answer to the above mentioned problems. They also show the possibility to measure very diluted samples at the GILDA beam-line. Experimental. The a-ge:h thin films were prepared by rf sputtering a c-ge target in an Ar+H 2 atmosphere. The doping was performed by co-sputtering small, solid pieces of the source material together with the Ge target. Details on samples preparation can be found in ref. [1-3]. The typical thickness of the film was 3 µm. The defect density was estimated from absorption coefficient (á 0.7 ) at 0.7 ev photon energy using the calibration constant reported in the literature [4]. The total dopant concentration was inferred from proton-induced X-ray emission. Samples of a-ge:h with dopant concentrations ranging from to cm -3 were examined. The average thickness of the films was of 3 µm with a number of doping atoms exposed to the x-ray beam (4 mm x 1.5 mm) of the order of 3*10 13 to 8* In order to determine the local order around the dopant specie a measurement of reference samples was carried out. In the case of gallium a gallium-implanted crystalline Ge sample with concentration of cm -3 was measured. For indium two reference samples were used: a crystalline germanium with impurity concentration of cm -3 indium atoms and a InAs sample. The difficulty in obtaining a crystalline sample doped with antimony induced us to use GaSb as reference. For the case of bismuth no crystalline reference sample was available. The K-edge of the doping elements (Ga, In, Sb, Bi) has been measured in fluorescence mode, at GILDA beam line (BM08), at room temperature, by a nitrogen-cooled pure Ge multi-detector. Due to the low concentration rates and the efficiency of the Ge detector, 15 to 20 seconds of acquisition time per point were used to record each spectrum. Results. Some typical fluorescence spectra are shown in fig.1. The EXAFS signal has been extracted using a technique already described elsewhere [5].

2 Fig_1. Fluorescence spectra relative to the a-ge:h film doped with Ga (left) and Sb with a concentration of cm -3 and cm -3 respectively. The dotted line (left) refers to the first degree polynomial used for the pre-edge removal while the inset in the right figure represent the k 2 weighted EXAFS signal. In figure 2a is reported, as an example representative of all the measurements, the k-weighted EXAFS signal relative to the gallium-doped samples [9] and the crystalline reference sample. In the same figure (b) the corresponding Fourier transform has been reported. The signal-to-noise ratio is satisfactory even at the smallest impurity concentration (~ cm -3, corresponding to ~ 0.003% at. %) even though the extension in the energy range is shorter in comparison with that of the more doped samples. The Fourier transform indicates that, as expected, the peaks relative to the second and third shell are absent in all the amorphous samples. A different situation attains the Bi doped samples (figure 3), for which no reference sample was available for comparison. In that case a theoretical approach was applied. A germanium crystal structure of 87 atoms with a central Bi atom in a Ge site, was simulated by the FEFF code: the theoretical signal obtained in such a way constitued the base for the comparison with the experimental ones. To extract the information on the local structure around the dopant specie, the EXAFS signal of the X-Ge first shell were Fourier back-transformed (X refers to the Ga, In, Sb and Bi). The information regarding the co-ordination number and the local order around the impurity atom of the investigated sample were extracted, except for bismuth, by a reference signal obtained from a crystal. As an example of this procedure, coordination number and Äó 2 (sample-model Debye-Waller difference) are summarised for three dopant species in figure 4. It is apparent from such figure that a similar behaviour of the coordination number as a function of impurity concentration can be deduced (solid symbols). The effect of rising doping concentration is a general lowering in coordination of the dopant atom. For Ga and In a four-fold coordination is apparent while for Sb a slightly less value is reported at the lowest concentrations. Values close to 2 are relative to the highest concentrations for all the impurity atoms. An opposite behavior is depicted in figure 5 for the bismuth-doped sample. In that case an increase in coordination with rising the dopant concentration was evidenced. It is interesting to note that at whatever concentration the bismuth-doped samples all were characterised by a poor conductivity [11]. The abrupt decrease in co-ordination doesn't follow a model proposed by Street [6] ( the modified 8-N Mott's rule ) which was well attained for P and B impurities in a-si:h. Beside, only 1% of four-fold coordinated sites was found electrically active. Moreover, the comparison between conductivity and coordination shows only a partial agreement, from which one can deduce that the conduction mechanism is not completely related to the substitutional doping.

3 Figure2. a) Extracted EXAFS signal multiplied by k relative to the Ga doped samples in order of increasing concentration (the most diluted is in the lowermost figure) and the crystalline reference sample (the uppermost figure). The signal-to-noise ratio for a concentration of ~ at.% is satisfactory. b) The corresponding Fourier transform. The same Hanning window was applied to all the data before the transformation. Fig_3. Bismuth doped sample. In the figure is reported (dotted line) the FT of the experimental signal. The continuos red line is the best-fit obtained by FEFFIT code. The reference signal was obtained in a theoretical way by FEFF.

4 Fig_4 Co-ordination number and Debye-Waller factor difference between sample and model for three measured dopant species. For an estimation of the disorder induced by the doping process was used a comparison method, with a crystal as reference sample, based on the cumulant expansion where the second cumulant represent the Debye-Waller factor. By this method it is possible to estimate the variation of the second cumulant which is related to the overall disorder of the sample under investigation with respect to the reference model. The outcomes of this investigation show that hydrogen can reduce the effect of the doping-induced network stress by completing dangling bonds [7]. The adding of chemical impurities surprisingly reduce further the disorder as can be seen in the fig. 4 (open symbols). A completely different outcome is related to the case of bismuth [10]. For this dopant, the very low coordination number shown at the lowest doping level, means that Bi does not fit naturally neither in 3-fold nor in 4-fold co-ordination because of its atomic size, incompatible with a stretching in the hosting network. This is confirmed by the evidence that, with increasing concentration, the DW factor has a growing trend: the value of (2.5±0.5)x10-3 Å 2 found for the sample having 1.1x10 19 cm -3 Bi atoms increases up to (6.5±0.5)x10-3 Å 2 for the most doped sample. This noticeably high value clearly indicates that Bi sites are much less ordered than the other column V element (Sb) sites. The results on local order and coordination of Ga,In,Sb and Bi impurities are consistent with data on electronic transport. The changes of the electrical conductivity induced by Bi impurity in a-ge:h films were studied by Burmeister [11] and indicate that Bi is a poor doping element even at the

5 higher concentrations. The room-temperature conductivity changes only by one order of magnitude for Bi concentration ranging from 4x10 17 to 2.5x10 20 cm -3. The activation energy changes by as little as 0.1 ev in the whole range and the defect density induced by doping remains much smaller than for the other column V elements. Figure5. Co-ordination and Debye-Waller factor for the bismuth doped a-ge(h) sample. The behavior of this two paramters is odd with respect to the other dopants. On the basis of these first results we proposed a new model to explain the doping properties of group III metals in a-ge:h [8]. The mechanism proposed is not related to the fraction of four-fold co-ordinated doping atoms only, but it considers the compressive stress induced by the atomic and ionic size differences between the germanium atom and the four-fold co-ordinated impurity and its effect on the co-ordination and electrical conductivity. References: 1. R. Zanatta and I. Chambouleyron, Phys. Rev. B 46, 2119 (1992) 2. F. Fajardo and I. Chambouleyron, Phys. Rev. 52, 4965 (1995) 3. D. Comedi,F. Fajardo and I. Chambouleyron, Phys. Rev. 52, 4974 (1995) 4. C.F.O. Graeff, M. Stutzmann and K. Eberhard, Philos. Mag. B 69,387 (1994) 5. G. Dalba, P. Fornasini, M. Grazioli and F. Rocca, Phys. Rev. B 52, (1995) 6. R. A. Street, Hydrogenated amorphous silicon 1991 (Cambridge Univ. Press, UK) 7. G. Dalba, P. Fornasini, R. Grisenti, I. Chambouleyron and C.F.O. Graeff (1994) J. Phys.: Condens. Matter Chambouleyron, D. Comedi, G. Dalba, P. Fornasini, R. Grisenti and F. Rocca (submitted to Phys. Rev. Lett.) 9. G. Dalba, P. Fornasini, R. Grisenti, F. Rocca, D. Comedi and I. Chambouleyron, Appl. Phys. Lett. 74 2, 281 (1999)

6 10. G. Dalba, P. Fornasini, R. Grisenti, F. Rocca and I. Chambouleyron, Appl.Phys. Lett. 81 4, 625 (2002) 11. F. Burmeister, MSc Thesis, Upsala Univ. and UNICAMP (1998), unpublished Publications: Dalba G., Fornasini P., Grisenti R., Rocca F., Chambouleyron I., "Local order of Sb and Bi dopants in hydrogenated amorphous germanium thin films studied by extended x-ray absorption fine structure". App. Phys. Lett., 2002, Vol. 81, pp Chambouleyron I., Comedi D., Dalba G., Fornasini P., Grisenti R., Rocca F., "Local coordination and electronic doping of column III metals in hydrogenated amorphous germanium.". Journal of non-crystalline solids 2000, Vol , pp Chambouleyron I., Comedi D., Dalba G., Fornasini P., Grisenti R., Rocca F., "Internal stress-induced changes of impurity coordination and doping mechanism in a-ge:h doped with column III metals". Solid state comm., 2000, Vol. 115, pp Dalba G., Fornasini P., Grisenti R., Rocca F., Comedi D., Chambouleyron I., "Local coordination of Ga impurity in hydrogenated amorphous germanium studied by extended x-ray absorption fine-structure spectroscopy". App. Phys. Lett., 1999, Vol. 74, n. 2, pp Conferences: I. Chambouleyron, D. Comedi, G. Dalba, P.Fornasini, R. Grisenti, and F. Rocca Local Coordination and Electronic Doping of Column III Metals in Hydrogenated Amorphous Germanium" ICAMS 18 - International Conference on Amorphous and Crystalline Semiconductors, Utah (USA) Grisenti R., Dalba G., Fornasini P., Rocca F., Chambouleyron I., Comedi D., "EXAFS determination of the local environment of column III and V elememts in amorphous germanium thin films.". Contributo a INFMeeting, Genova, 12/16 giugno Grisenti R., Dalba G., Fornasini P., Rocca F., Chambouleyron I., Comedi D., "Investigazione sull'ordine locale di droganti delle colonne III e V in film di germanio amorfo idrogenato". Contributo a Primo Convegno Utenti GILDA, Folgaria (Trento), gennaio 2001.

7 R. Grisenti,G. Dalba, P. Fornasini, F. Rocca and I. Chambouleyron "Local order around Sb and Bi dopants in hydrogenated amorphous germanium thin films." ICAMS 19 - International Conference on Amorphous and Crystalline Semiconductors, Nice (France) August Experiment n. HS 536 Experiment n Experiment n Experiment n