We present a new general formalism for investigating the second-order optical response of solids, and illustrate it by deriving expressions for the second-order susceptibility tensor 2 (Ϫ ⌺ ; ␤ , ␥), where ⌺ ϭ ␤ ϩ ␥ , for clean, cold semiconductors in the independent particle approximation. Based on the identification of a polarization operator P that would be valid even in a more complicated many-body treatment, the approach avoids apparent, unphysical divergences of the nonlinear optical response at zero frequency that sometimes plague such calculations. As a result, it allows for a careful examination of actual divergences associated with physical phenomena that have been studied before, but not in the context of nonlinear optics. These are ͑i͒ a coherent current control effect called ''injection current,'' or ''circular photocurrent,'' and ͑ii͒ photocurrent due to the shift of the center of electron charge in noncentrosymmetric materials in the process of optical excitation, called ''shift current.'' The expressions we present are amenable for numerical calculations, and we demonstrate this by performing a full band-structure calculation of the shift current coefficient for GaAs.

We show that, when fundamental optical beams are present in a noncentrosymmetric medium simultaneously with their sum-frequency beam, quantum interference between single-and two-photon transitions modifies the net absorption, if the sum frequency corresponds to an energy greater than the band gap. At a macroscopic level this effect can be related to the imaginary part of a second-order susceptibility and can be used to coherently control carrier populations and optical absorption. We illustrate this novel effect using phased 1550 and 775 nm, 120 fs pulses incident on GaAs at 295 K. PACS numbers: 78.47.+ p, 78.55.Cr The field of nonlinear laser optics dates from the observation of second-harmonic generation in crystalline quartz [1]. Although a myriad of nonlinear optical processes have been discovered since then [2-4], second-order ͑x 2 ͒ processes such as sum-and difference-frequency mixing continue to capture most of the interest. Of necessity x 2 is nonzero only in noncentrosymmetric media, such as certain crystals. When such a crystal is transparent at all frequencies involved, x 2 is real and the crystal can act as an optical "catalyst" to convert incident energy into new optical frequencies. If, however, the crystal is absorbing for at least one of the frequencies, the imaginary component of x 2 ͑Im͓x 2 ͔͒ is nonzero. In the case of sum-frequency generation, it is generally believed [5] that Im͓x 2 ͔ plays no role in energy absorption. Here we demonstrate that when coherent fundamental and sum-frequency beams are simultaneously present in a crystal, and if the sum frequency falls in a region of band absorption, Im͓x 2 ͔ can contribute to the removal of energy from all beams. This is due to the interference of single-and two-photon absorption pathways. This effect can be used in an active sense to coherently control optical transmission and band populations through the relative phase of the fields. We experimentally demonstrate this new effect in bulk GaAs at 295 K, and explicitly discriminate its origin and nature from the coherent control of current injection in semiconductors [6].

The reflectance anisotropy has been calculated by microscopic tight-binding theory for various configurations of the As-rich GaAs(100) c͑4 3 4͒ and ͑2 3 4͒ reconstructions, based on precise atomic coordinates from ab initio total-energy minimization. The comparison to experimental reflectance anisotropy in combination with scanning tunneling microscopy and low energy electron diffraction allows one to identify precise correlations between structural units and optical features. Optical spectroscopy has become an important tool of surface analysis in the last years, due to its high sensitivity and in situ applicability [1]. In particular, reflectance anisotropy spectroscopy (RAS) is increasingly used for monitoring the growth of epitaxial structures in molecular beam epitaxy (MBE) or in metal organic vapor pressure epitaxy (MOVPE) [1][2][3][4]. However, theoretical understanding is needed in order to fully exploit its potential.Among the technologically important (100) surfaces of III-V semiconductors, the most intensively studied "prototype" is GaAs(100). A variety of different reconstructions, dependent on surface stoichiometry, exist; the three main reconstructions are the As-rich c͑4 3 4͒, the Asrich ͑2 3 4͒͞c͑2 3 8͒, and the Ga-rich ͑4 3 2͒͞c͑8 3 2͒ phases [2][3][4][5][6]. Several structural models of the As-rich phases are discussed in the literature [7-9] (see Fig. 3): At high As coverage the c͑4 3 4͒ phase should consist of three top As dimers bonded to the next complete As monolayer. Annealing up to around 400 ± C leads to the ͑2 3 4͒͞c͑2 3 8͒ phase. Total energy calculations [7][8][9] predict two different stable ͑2 3 4͒ geometries depending on the preparation conditions: the so-called b2 and the a structure. The b2, containing two top As dimers and one As dimer in the exposed third layer, accounts for the ͑2 3 4͒͞c͑2 3 8͒ phase. The so-called a structure is characterized by Ga dimers in the second layer besides the two top As dimers.The local atomic structure of the surface has often been claimed to play a key role in determining the surface optical anisotropy [1][2][3][4][5]. Hence, using a reliable theoretical description it should be possible to relate the optical response to the atomic surface structure. At present, few examples of good agreement between experiments and calculations of the optical response based on the one-electron band structure approximation, employing semiempirical tight-binding [10][11][12] as well as ab initio plane-wave expansions [13], are available. For GaAs(100), however, theoretical results are rather unsatisfactory [14,15], and recently it was hypothesized that RAS line shapes in many cases may have little to do with the atomic structure of the surface, but are rather determined by surface-induced changes of excitonic and local-field effects on bulk transitions [16,17] which are not included in the above mentioned calculations.In this work, we demonstrate that calculations based on the one electron band structure approximation indeed yield a good description of experiment...

No abstract

We show that ballistic current generation can occur in a semiconductor via quantum interference between absorption pathways for orthogonal polarization components of a single-frequency beam. This effect occurs for a subset of noncentrosymmetric materials, is macroscopically associated with a second-order nonlinear optical susceptibility, and produces current injection linearly proportional to the beam intensity. We demonstrate this in wurtzite CdSe (Eg=1.75 eV) at 295 K using cw and femtosecond optical sources of wavelength 600–750 nm (2.07–1.66 eV). The intensity and spectral dependence are in reasonable agreement with a first-principles calculation. Continuous current density of 30 μA cm−2 is produced for 60 mW cm−2 intensity at 633 nm.