We present experimental evidence of the equilibrium coexistence between crystalline phases in heteroepitaxial films of MnAs on GaAs. The phases, which can coexist in the bulk system only at one temperature point, coexist in the epitaxial film over a wide temperature interval. An apparent contradiction with the Gibbs phase rule is resolved by the presence of strain in the film. The Gibbs phase rule limits the coexistence between phases with the same chemical composition to a single temperature.

more »

... e present experimental evidence of an equilibrium phase coexistence in heteroepitaxial films over a wide temperature interval. In films of MnAs grown epitaxially on GaAs, we observe the coexistence of two structurally distinct phases, hexagonal aMnAs and orthorhombic bMnAs, in a range from the bulk phase transition temperature to at least 20 ± C below it. The fraction of the low-temperature phase decreases almost linearly when approaching the phase transition temperature. Thermal cycling does not reveal any hysteresis. An apparent contradiction with the Gibbs phase rule is resolved by the presence of long-range elastic interactions in the strained heteroepitaxial film. The epitaxial coupling of the film to the substrate does not allow a change of the film sizes (and, hence, the mean strain) in the plane of the interface. As a result, the minimum of the free energy of the film, which includes the elastic strain energies of both phases, is realized by a coexistence of these phases. MnAs on GaAs is a promising heteroepitaxial system which integrates magnetic and semiconductor properties. It has been intensively studied during the past years [1,2]. Figure 1 sketches the epitaxy of MnAs on a GaAs(001) surface. Below 40 ± C, the bulk MnAs crystal is ferromagnetic and forms the hexagonal aMnAs phase (structure of bulk phases and the phase diagram of MnAs are reviewed in Ref. [3]). In epitaxy, the (1100) side facet of the hexagonal prism is attached to the GaAs(001) surface. At approximately 40 ± C, the bulk aMnAs experiences a first-order phase transition to the paramagnetic orthorhombic phase bMnAs. Its unit cell is shown in Fig. 1 by the dashed line: the hexagon anisotropically shrinks in both directions, while the height of the prism (perpendicular to the plane of the figure) does not change. A further phase transition in bulk MnAs takes place at 125 ± C. This transition is continuous and results in the gMnAs phase, which is hexagonal again. The structure of bulk MnAs crystals [4], the epitaxial relationships of MnAs on GaAs [1,2], and the structure of the interface [2] are known. MnAs grows epitaxially on GaAs(001) despite a very large mismatch in the GaAs͓110͔ direction which amounts to 33%. The trans-mission electron microscopy studies [2] show that every sixth GaAs͕220͖ plane fits into every fourth MnAs͕0002͖ plane, which reduces the actual mismatch to 5%. This mismatch, as well as the mismatch along the perpendicular direction (7.7%), is released by regular arrays of misfit dislocations. The MnAs films reveal a unique epitaxial orientation on the polar GaAs(001) surface, namely, ͑1100͒ MnAs k ͑001͒ GaAs and ͓0001͔ MnAs k ͓110͔ GaAs, which was checked in the present study by in situ reflection high-energy electron diffraction. It is essential for the considerations below that the orientation of the film with respect to the substrate is unique. There are no rotationally equivalent domains (twins), albeit translational domains can be present. At the aMnAs-bMnAs transition, the unit cell shrinks in the plane of the interface, as shown in Fig. 1 , resulting in a strained film. Hypothetically, three scenarios are possible. If misfit dislocations can be generated to release the elastic strain energy, the phase transition would proceed exactly in the same way as in bulk MnAs. However, generation of misfit dislocations is hardly possible near room temperature. In the opposite case, if the dislocations could neither be generated nor moved by glide, the strain could not be released nor redistributed. It would be locally fixed, and the phase transition would proceed uniformly, albeit at a temperature different from the bulk transition temperature. The third possibility is realized when the dislocations cannot be generated because of the low temperature but the existing misfit dislocations can glide along the interface. Then, the strain can be redistributed, and only the mean strain over the whole sample is fixed. The free energy minimum of the film is reached through the coexistence of domains of two phases with different strain. The x-ray FIG. 1. Scheme of the epitaxy of MnAs on GaAs(001). 0031-9007͞00͞85(2)͞341(4)$15.00