Epitaxial liftoff is a post-growth process by which the active part of a semiconductor heterostructure, the epitaxial layer, is removed from its original substrate and deposited onto a new substrate. This is a well established technique in GaAs-based heterostructures where epitaxial liftoff can be achieved by exploiting the contrast in the etch rates of GaAs and AlAs in hydrofluoric acid. We report here successful epitaxial liftoff of a ZnSe-based heterostructure. We find that a metastable

more »

... of MgS acts as a perfect release layer based on the huge contrast in the etch rates of ZnSe and MgS in hydrochloric acid. Epitaxial liftoff of millimeter-sized ZnSe samples takes a fraction of the time required for GaAs liftoff. Photoluminescence experiments confirm that the liftoff layer has the same optical characteristics as the original wafer material. The development of heterostructures has revolutionized both the physics and the technology of semiconductors. Inherent to the technique is the growth of high quality semiconductor layers on a single crystal substrate. The substrate must be of high crystal quality with a lattice constant similar to that of the epitaxial layer itself. In some cases, these conditions impose unwelcome restrictions on the range of heterostructures that can be produced. Also, for some technologies, for instance solar cells, a single crystal substrate is simply too expensive. Epitaxial liftoff is a technique which can add flexibility to the fabrication of semiconductor devices. 1 It allows the epitaxial layer to be removed from its substrate in a post-growth processing step so that the epitaxial layer can be deposited onto a new substrate with different functionality. The liftoff step also allows the original substrate to be reused. Up until now, epitaxial liftoff has been achieved using the remarkable contrast in the etch rates of GaAs and Al x Ga 1−x As in HF. 2,3 For x ജ 0.5, the Al-containing alloy etches many orders of magnitude faster than GaAs. Epitaxial liftoff of a GaAs-based heterostructure can therefore be achieved by growing a thin sacrificial layer of AlAs between the substrate and the active layer. A sacrificial layer of thickness between 3 and 40 nm is effective. 4 After growth a wafer piece is submerged in HF, the HF etching through the AlAs layer but having a negligible effect on GaAs, InGaAs, and low Al-content AlGaAs alloys. In addition to the etch selectivity, a crucial idea is to stress the epitaxial layer so that as the etching proceeds, the epitaxial layer curls up slightly. 3 This is important as otherwise the etching would proceed too slowly and eventually stop altogether. The stress can be applied by several methods, the simplest of which is to anneal a layer of wax. 2,3 This idea has been used to lift crack-free layers as thin as 20 nm from the original substrate. 5 This technology has been exploited in a number of semiconductor devices. A laser diode, 6,7 a solar cell, 8 rib waveguides, 9 and a photodetector 10 have all been transferred onto a foreign substrate. Attempts have been made to speed up the process, and to make it compatible with industrial scale semiconductor processing in order that epitaxial liftoff also becomes a commercially viable technology. 1, 4 Despite recent progress, epitaxial liftoff remains a possibility only for GaAs-based heterostructures. While InP-based and ZnSe-based layers have been lifted from their original substrates, 4,11 in both cases the technology depends on an AlAs release layer. In the case of II-VI materials, this is inconvenient as it not only increases the thickness of the epitaxial layer which can be lifted off but it also requires both III-V and II-VI molecular beam epitaxy. The absence of a viable technique for carrying out II-VI epitaxial liftoff is a serious limitation in at least two areas. First, while Bragg stacks can be grown in a GaAs-based heterostructure by exploiting the difference in refractive index between the latticematched pair, GaAs and AlAs, there is no lattice-matched pair of II-VI materials with sufficiently different refractive indices. Epitaxial liftoff would provide a natural way in which thin high quality ZnSe layers could be transferred onto a conventional dielectric Bragg mirror. This is important for polariton parametric amplification, 12,13 a very attractive concept for ZnSe quantum wells where excitons can be stable up to high temperatures. 14 Second, magnetic atoms can be incorporated into II-VI compounds without introducing any doping and this is driving the interest in II-VI materials for spintronics applications. 15 Epitaxial liftoff could offer valuable advantages in combining this property with ferromagnetic layers. 16 We present a method of performing epitaxial liftoff of ZnSe-based heterostructures. We have discovered that a thin layer of MgS in a ZnSe-based heterostructure acts in an analogous way to AlAs in a GaAs-based heterostructure. Bulk MgS exists in the rock-salt crystal structure but we are able to grow layers up to 134 nm thick in the zinc-blende structure on a GaAs substrate. 17 Up to this thickness, the layer quality is very high as can be judged from the excellent optical properties of ZnSe/ MgS quantum wells. 14 At larger thicknesses, the MgS layer reverts back to the rock-salt structure and further zinc-blende growth is clearly of low quality. We exploit here the pronounced difference in chemical properties between MgS and ZnSe to perform epitaxial liftoff. Remarkably, we find that the etch rate of a MgS layer is ϳ10 8 times larger than the etch rate of a ZnSe layer. a) Electronic mail: a.curran@hw.ac.uk APPLIED PHYSICS LETTERS 86, 011915 (2005)