This paper reports the experimentally observed change of exciton-polariton eigenenergies near the surface of a semiconductor. Normal-incidence-reflection spectra and attenuated-totalreflection (ATR) spectra are measured in the n =1 exciton-polariton energy region. It is shown that ATR spectra probe regions near the surface whereas reflection spectra probe more deeply into the crystal bulk. Model calculations"which include spatial dispersion and depth-dependent eigenenergies and damping of excitonic polaritons, yield excellent agreement with experiments for various semiconductors. This agreement proves that reflection spectra are determined not only by bulk properties of excitonic polaritons, but reveal also the properties of the transition region at the crystal surface. Therefore, information extracted so far from reflection spectra about additional boundary conditions of excitons and exciton-free surface layers may have to be revised.

No abstract

Surface exciton polaritons are observed in a semiconductor for the first time. The modes are excited in ZnO crystals by the method of attenuated total reflection. Changing the angle of incidence of the ultraviolet radiation gives an experimental dispersion relation for surface exciton polaritons.Surface excitons have been treated theoretically by several authors, 1 " 5 Polariton dispersion relations and the reflectivity of crystals for the attenuated-total-reflection (ATR) method have been calculated. 4 ' 2 The surface exciton frequencies always lie between the transverse and longitudinal resonance frequencies oo T and oo L of the bulk excitons for a given wave vector. Because of spatial dispersion, surface excitons have among surface excitations the unique property of coexistence with a bulk mode in the energy region between oo T and ix) L . Consequently, energy transfer between these modes is possible. 3 Surface modes can be excited by an ATR arrangement. This method uses a prism to couple electromagnetic waves of wave vector k > oo/c vac across a thin spatial gap with the sample surface. The ATR method has been applied successfully to detect surface phonon polaritons and surface plasmons in various media. 6 Extending ATR to the shorter wavelengths of surface excitons is difficult because of the stringent requirements on control of the spacing between crystal and prism and because of exciton-free surface layers. 4 ' 7 According to calculations by Maradudin and Mills, 2 the reflectivity changes &R/R in an ATR experiment are expected to be of the order of 10 " 5 . All these reasons may account for the fact that experiments on surface excitons have not been reported until now. However, a direct observation of surface excitons would offer new possibilities for obtaining information about exciton-free surface layers and spatial dispersion.In this paper we report the first optical excitation of surface exciton polaritons in a semiconductor. We used an arrangement of attenuated total reflection at helium temperatures. Our experimental results give the dispersion relation of surface exciton polaritons in the short-wavelength region near the band gap of ZnO crystals.A careful choice of the crystal material is necessary to obtain the best conditions for the experimental excitation and observation of surface excitons. It is known for some materials that image charges or field-induced ionization of excitons creates an exciton-free surface layer at least 1 to 2 exciton Bohr radii thick. 7 ' 8 The exciton-free surface layer has to be as thin as possible in order to excite excitons by an ATR technique which uses the evanescent waves "leaking out" of a prism while light is totally reflected inside. Only for thin exciton-free surface layers are the exponentially decaying electromagnetic waves still sufficiently intense at the crystal regions where excitons exist. From the free-exciton reflectance spectra it is known that the exciton-free surface layer of ZnO is less than about 30 A, 9 probably the smallest one of the II-VI semi...

The well-known exciton reflectance spectra of ZnO a t helium temperatures are changed drastically by surface layers. To explain the experimentally observed structure the reflectivity of a medium with harmonic oscillators (nonspatial dispersion) or with spatial dispersion and with an additional surface layer is calculated. The surface layer is either an intrinsic exciton-free layer or an impurity layer in which multiple reflections and interference take place. From the comparison of experiment and calculation the following results are obtained: firstly, spatial dispersion of free excitons in ZnO has t o be considered, secondly, the lineshapes of the closely spaced A1 and B1 excitons are changed in a very different way, thirdly, the numerical evaluation yields the longitudinal resonance energies of the B1 and C1 excitons. By heavily doping a surface layer an enlarged exciton-free layer could be produced which is demonstrated by the reflectance spectra.Die bekannten Reflexionsspektren freier Exzitonen in ZnO bei Heliumtemperatur werden durch Oberflachenschichten drastisch verandert. Urn die experiment.el1 beobachtete Strukt u r zu erklaren, wird das Reflexionsvermogen eines Mediums mit harmonischen Oszillatoren (nichtraumliche Dispersion) oder mit raumlicher Dispersion und mit einer zusktzlichen Oberflachenschicht berechnet. Die Oberflachenschicht ist entweder eine intrinsische exzitonenfreie Schicht oder eine Fremdschicht, in der es zu Vielfachreflexionen und Interferenzen kommt. Aus dem Vergleich von Experiment und Rechnung werden die folgenden Ergebnisse erhalten: erstens hat man bei den freien Exzitonen in ZnO raumliche Dispersion zu beriicksichtigen, zweitens werden die Linienformen der eng benachbarten A1 und B1 Exzitonen sehr verschieden beeinflurjt,, drittens liefert die numerische Auswertung die longitudinalen Eigenfrequenzen der B1und Cl-Exzitonen. Durch starkes Dotieren einer Oberflachenschicht kann eine vergrorjerte exzitonenfreie Schicht erzeugt werden, was sich in den Reflexionsspektren zeigt.