Structural and Electronic Properties of In-Situ Phosphorous-Doped Ge Layers Grown by Reduced Pressure – Chemical Vapour Deposition

Direct gap photoluminescence, electroluminescence and optical gains at wavelengths close to 1.6 µm have recently been demonstrated in tensile-strained, phosphorous-doped Ge layers grown on Si(001). The electrons given by the n-type impurities to the Ge crystalline lattice occupy the L valleys of the conduction band, exalting fast, direct transitions between the Γ point of the conduction band and the heavy and light holes sub-bands of the valence band. There is however a scarcity of results on the impact of PH₃ on the structural and electronic properties of Ge:P layers (with GeH₄ as the Ge gaseous precursor). We have quite conclusively shown in J. Cryst. Growth 347, 37 (2012) that low deposition temperatures and high growth pressures yielded the high P contents aimed for in Ge (> 10¹⁹ cm^-3; suppression of P surface segregation).

We have thus grown at 400°C, 100 Torr on p-type Si(001) substrates thin (~ 350 nm) and thick (~ 700 nm) Ge:P layers with various PH₃ flows. Those Ge:P layers were submitted afterwards to a 60s, 750°C anneal or to a 3x(750°C, 10s / 875°C, 10s) thermal cycling, both under H₂, to cure crystalline defects and improve electronic properties. We will quantify in the following the impact of PH₃flow, thickness and annealing scheme on the structural, electrical and optical properties of the resulting Ge:P layers.

A transition from smooth, “mirror-like” to rough, “milky” surfaces was evidenced above a F(PH₃)/F(GeH₄) mass-flow ratio close to 10^-3 (irrespectively of the thickness or the H₂ anneal used). This was associated with a sudden drop in deposited thickness (from 739 down to 579 nm and from 365 down to 303 nm, respectively). Hills together with dislocation slip lines running along the <110> directions (for thick, cyclically annealed layers), were evidenced in Atomic Force Microscopy. The surface root mean square roughness and Z range increased almost exponentially with the PH₃ flow, reaching really high values above the threshold PH₃flow (33 and 242 nm for thick Ge:P layers <=> 7 and 59 nm for thin Ge:P layers). Thick Ge:P layers were otherwise 70% rougher with a single 750°C anneal than with a cyclic anneal. The opposite was true with thin Ge:P layers, which were 15% smoother with the former than the latter.

X-Ray Diffraction showed that all Ge:P layers were slightly tensile-strained and diffracted coherently irrespectively of the process conditions used. The strain was however slightly higher after a cyclic anneal than after a 750°C, 60s anneal (macroscopic degree of strain relaxation R: 105.5 – 106% <=> 104 – 104.5%). Four points probe resistivity measurements showed that the resistivity was higher (and thus the P⁺ ions concentration lower) (i) in thin than in thick Ge:P layers and (ii) after a cyclic anneal (up to 875°C) than after a straightforward 750°C, 60s anneal (P out-diffusion). P⁺ ions concentrations more than 2x10¹⁹ cm^-3were nevertheless achieved.

Finally, room temperature photoluminescence measurements showed a factor close to 10 increase of the PL peak Intensity together with a red shift in Ge:P layers (figure). The PL intensity dropped dramatically irrespectively of the process conditions when F(PH₃)/F(GeH₄) significantly exceeded the threshold value (rough but still crystalline layers). Higher PL peak intensities were otherwise obtained after 750°C, 60s anneals than after cyclic anneals (which yield higher crystalline quality, smoother and slightly more tensile-strained layers); this is likely due to P loss because of out-diffusion.

Using such Ge:P layers as the core of micro-bridges (Nature Photonics 7, 466 (2013) could be interesting as high strain and in-situ P doping would then be combined, with potentially a beneficial impact on optical properties.

The support of the “Phare” DSM/DRT Project is gratefully acknowledged.