Type I Band Alignment inSi1−xGex/Si(001<

Abstract

We present experimental verification of a type I conduction band alignment for coherently strained Si~,Ge, layers in (001) silicon, with 0.15~x~0.38. A novel substrate bending scheme is used to apply in-plane uniaxial compressive and tensile stress along the [100] and [110] directions. Band edge photoluminescence from SiGe and Si is shifted with stress in accordance with deformation potential theory. Tensile stress along [110]allows clear distinction between types I and II band alignment where the predicted shifts are in opposite directions. PACS numbers: 62.20.Dc, 68.55. -a, 71.35.+z, 73.20.Dx Si& Ge alloys deposited epitaxially and coherently on Si(001) substrates have been extensively investigated in recent years [1 -14] and are rapidly gaining technological importance. Photoluminescence (PL) provides accurate measurement of the SiGe band gap in unstrained (cubic) alloys [1] and strained SiGe [2] epitaxially deposited on (001) Si. Quantum confinement effects [3] and SiGe/Si interface integrity [4] have also been studied by PL. However, evidence of the band lineup at the SiGe/Si heterointerface has remained elusive. Although it is accepted that the valence band (VB) offset AE is much larger than AE, for the conduction band (CB), more precise knowledge of band alignments is essential to optimize SiGe heterostructures for advanced devices. Previous attempts at estimating band alignments in this system have met with only modest success [5 -13]. Controversy even remains as to the sign of AE"and thus the extent of the VB discontinuity is in dispute. Early theoretical studies [5] suggested type I (electron confinement in SiGe) alignment (maximum AE, = 20 meV at x -0.2). More recent theory [7] supports type II (electron confinement in Si) behavior with AE,~x. Recent PL measurements of SiGe quantum wells (QW's) show both type I [8,9] and type II [10] behavior, whereas earlier luminescence studies [2,11] were interpreted to support either band alignment. Northrup et al. [12], using PL under hydrostatic pressure, claimed "circumstantial" evidence for type I band alignment for x -0.25. Mantz et al. [13], using uniaxial [001] stress, present PL data implicitly supportive of type I behavior at zero stress, in contradiction with their recent publication [10] in which type II behavior is claimed. Much of this confusion is due to the absence of a technique to uniquely identify the electron and hole states responsible for the observed optical transitions. Here we describe a novel technique providing in-plane uniaxial tension and compression along [110]and [100]. This reduces the SiGe QW lattice symmetry from tetragonal to orthorhombic while simultaneously distorting intermediate Si layers into tetragonal symmetry. The degeneracy at the band edges can be lifted systematically, unambiguously revealing the CB's and VB's contributing to the observed PL transition. External [100] and [110]uniaxial stresses were applied during PL by bending a Si wafer to create an elastic half space on either side of its center plane. This results ...

Type II Band Alignment inSi1−xGex/Si(001)Thewalt1, Harrison2, Reinhart3 et al. 1997Phys. Rev. Lett.1194530View full textAdd to dashboardCite

“…For both types of stress, the high excitation data shows good agreement with the blueshift calculated for type I transitions, which also is in agreement with the results reported in Refs. [16] and [17] and is consistent with the relatively high excitation levels used in those earlier studies. However, as the excitation level, and thus the charge accumulation, is lowered, we observe a strong deviation from type I behavior above some characteristic stress which decreases with decreasing excitation.…”

“…However, studies of the effects of compressive ͓001͔ uniaxial stresses by the same group produced results which were only consistent with type I alignment, although that conclusion was not explicitly stated [16]. More recently Houghton et al [17] extended that approach and, based on Si 12x2y Ge x ͞Si͑001͒ QW PL shifts under applied ͓110͔ and ͓100͔ uniaxial stresses, concluded that the band alignment was indeed type I. A surprisingly large CB discontinuity of at least 10 meV for x 0.15 was quoted, based on the absence of any indication of a turnover to type II PL up to the highest ͓110͔ tensile stresses studied.…”

“…The (001) biaxial tensile stress experiments used a novel wafer bending apparatus [20]. To facilitate direct comparison with the earlier findings of Houghton et al [17], PL measurements under ͓110͔ uniaxial tensile stress were also performed using their linear bending method.…”

Deep-level transient spectroscopy study of theEcenter inn−Siand partially relaxedn−Mamor1, Elzain2, Bouziane3 et al. 2008Phys. Rev. B7350View full textAdd to dashboardCite

“…Our partially relaxed Si 0.9 Ge 0.1 ͑ Ϸ 22% ͒ has a remaining strain of 0.33% and its band gap is about 45 meV below that of fully relaxed Si 0.9 Ge 0.1 . Based on previous results showing that the band gap shrinkage results in an upward shift of the valence band, 27,28 also taking into account that the valence band offset is for at least 90% of the band gap difference between Si and strained Si 1−x Ge x as well as between the strained Si 1−x Ge x and the relaxed Si 1−x Ge x for a given Ge content, 27 and finally, setting the valence band of Si equal to the energy reference, the resulting band gap variation diagram as a function of degree of strain in Si 0.9 Ge 0.1 can be deduced and is shown in Fig. 3.…”

Photocurrent of SiGe/Si Strained Multiple Quantum-Wells Grown by UHV-CVD

“…Strained Si/SiGe MQWs have been generally accepted that holes in these heterostructures experience potential wells in the SiGe layer and barriers in Si. The situation in conduction band (CB), however, is rather obscure under the influence of alloying and strain effects, resulting in some uncertainty on whether type-I or type-II band alignment is formed in a given multiple quantum well (MQW) structure (14)(15)(16)(17). However, being able to precisely determine the type of alignment has a direct impact on device properties, e.g., type-I alignment offers a larger band-to-band oscillator strength than type II; thus a stronger radiative emission efficiency is expected from type-I devices.…”