Infrared reflection of ion-implanted GaAs

Abstract

Gallium arsenide has been implanted with nitrogen ions at 1–3 MeV and fluences between 3.3×1013 and 2.0×1017 ions/cm2. Room-temperature infrared spectra are presented which show major changes in the reststrahl region. A sample having the highest fluence was isochronally annealed with 2-h 100°C steps from 200 to 600°C, and the implantation-induced reflectivity changes are annealed by 600°C. Changes of the dispersion parameters were determined by Kramers-Kronig analysis as well as by curve fitting with classical dispersion (CD) analysis. Examples of dispersion parameters obtained by different analyses are compared and their validity is discussed. The data of nonimplanted and samples implanted with fluence up to 3.3×1014 ions/cm2 can be satisfactorily fitted with CD analysis assuming the material to be optically homogeneous. With this assumption, attempts to fit the data of samples with fluences ≥3.3 ×1015 ions/cm2 were unsuccessful. By extending the CD analysis to a layer model a reasonable fit was achieved. The analyses indicate changes in the transverse- and longitudinal-optic-mode frequencies. The latter changes were confirmed by Raman scattering measurements. Both the analyses and interference fringes observed in the near-infrared reflection spectra indicate some implantation-induced increase in the high-frequency dielectric constant.

The classical oscillator model and dielectric constants extracted from infrared reflectivity measurements

“…The introduction of, or an increase in, the phonon damping parameter c p has negligible effect outside the reflectivity band, but reduces the reflectivity inside the band, with the effect being more pronounced on the high frequency side of the band. This trend was already illustrated theoretically by Picus et al [24] and can be seen in the experimental spectra of nitrogen implanted GaAs of Kachare et al [25], where the phonon damping increased as a function of dose. In subsequent calculations the value of c p = 1.6 cm À1 given in Table 1 will be used.…”

“…The carrier concentration was found to decrease and the carrier damping was found to increase as a result of polishing damage. Holm et al [16] refer to a report that the maximum shift in the phonon frequencies due to strain or damage is 2 cm À1 , which exceeds the change reported by Kachare et al [25]. In more recent results from Moore and Holm [9], the difference in the fitted optical phonon frequencies between semi-insulating undamaged samples differing only in thickness is 2 cm À1 or more.…”

“…The optical phonon frequencies and high and low frequency dielectric constants are then generally considered to depend on the type of material. Despite this, Kachare et al [25] reported significant changes in the material properties of ionimplanted undoped GaAs, in particular a decrease in x l and an increase in e 1 (together with the expected increase in c t ) as a function of dose. The reported [34] value of e 1 for amorphous GaAs is 12.25 ± 0.7, which is significantly larger than the value for crystalline GaAs given in Table 1, indicating that an increase in this parameter is possible.…”

“…Although Patrini et al [20] reported a value for e 1 lower than the accepted value for an epitaxial layer modeled as a semi-infinite medium, and the results of Kachare et al [25] for fits of the reflectivity of a low dose implanted GaAs sample using both models also gave a lower value for the static dielectric constant when the semi-infinite model was used, the differences from the benchmark values in these reports are much less and the results of Engelbrecht and Botha [4] appear unrealistic. Whereas the GaSb reflectance spectrum which they calculated from published refractive index data had a high frequency (beyond the phonon band) reflectivity in the order of 30%, the measured reflectivities of the epitaxial layers 6 and 9 lm thick in the same frequency range were of the order of 10%.…”

“…These must either be included in the model for the reflectivity [22], or the back surface of the sample can be roughened to prevent multiple reflections [29]. A layer-on-a-substrate model has been used for the reflectivity of implanted material [25,28], for simulating surface damage due to polishing [16,30], for simulating a surface depletion layer [16] and for epitaxial layers [4,16]. Multiple thin films have also been investigated [19,31].…”