1049
Cleaning of InGaAs and InP Layers for Nanoelectronics and Photonics Contact Technology Applications

Wednesday, October 14, 2015: 10:45
104-A (Phoenix Convention Center)
P. Rodriguez (CEA LETI), L. Toselli, E. Ghegin (CEA LETI, STMicroelectronics), M. Rebaud, N. Rochat, N. Chevalier, E. Martinez (CEA LETI), and F. Nemouchi (CEA LETI)
In the context of III-V contact technology for nanoelectronic and photonic applications and for a sake of reducing parasitic resistances between metal and source / drain regions, very cleaned and native oxide free surfaces must be provided prior to metal deposition. In order to meet the requirements set out in the ITRS roadmap and to propose a contact technology on III-V compounds compatible with the Si CMOS technology, we assume that the cleaning of S/D regions should involve a two-step procedure where a wet cleaning treatment is coupled with an in situ plasma exposure (e.g. a plasma treatment realised, without air break, in the same equipment as the metal deposition). If wet chemical cleanings of InGaAs and InP surfaces have already been reported in literature, in situ treatments have not been extensively considered.

In this work, we studied the pretreatment of InGaAs and InP layers by employing Ar- and He-based direct plasmas and NH3, H2, NF3/NH3 remote plasmas carried out in a 300 mm platform usually dedicated to silicide process and fully compatible with the Si CMOS technology. Moreover, the combination of wet chemical treatments followed by in situ plasma treatments have been investigated. The characterisation of cleaning efficiency has been performed on the CEA Minatec Nanocharacterisation Platform (PFNC) using several surface analyses like X-ray photoelectron spectroscopy (XPS), atomic force microscopy (AFM) and Fourier transform infrared spectroscopy in attenuated total reflection mode (FTIR-ATR).

FTIR-ATR results coupled with XPS analyses highlighted that Ar- and He-based direct plasmas were efficient for the removal of InGaAs native oxides whereas all the remote plasmas involved were inadequate to remove the III-V oxides. Moreover, for NF3/NH3 exposed samples, we noticed the addition of undesirable In-F and Ga-F bonds. Regarding Ar and He direct plasmas, investigations exhibited that both seem to be efficient for removing arsenic oxides whereas the elimination of indium and gallium oxides is more effective with Ar plasma. We also studied the addition of hydrogen into He direct plasma and, unexpectedly, we demonstrated that increasing the H2 content leads to a decrease in the removal of arsenic oxides. This trend could be correlated with the increase of pressure induced by hydrogen addition. On the other hand, the impact on indium and gallium oxides is not notable (i.e. the removal of group III oxides is nor enhanced, nor reduced) but we observed a reducing effect of hydrogen on indium and the emergence of In-In type bonds. Then, the combination of wet chemical treatment with in situ plasma was studied. We highlighted that the wet chemical treatment of InGaAs layers in concentrated HCl solutions is beneficial for reducing the oxide component of indium and gallium. Moreover, using a wet cleaning appears to modify the InGaAs layers behaviour towards plasma treatment. Indeed, He direct plasmas provide the best results for reducing As and In oxides whereas Ar plasmas are more convenient for eliminating Ga oxides. Finally, whatever the plasma pretreatment, no degradation of surface morphology and roughness was observed by AFM. The RMS values obtained after surface treatments are similar with the ones acquired for the reference samples.

InP surfaces appear to be more sensitive to air re-oxidation than InGaAs layers. For indium phosphide material, we showed that the surface was very sensitive to the direct plasma treatments. Indeed, AFM studies highlighted that the use of Ar-based plasma drastically impacts the surface morphology and its roughness. This impact on surface morphology is correlated with the increase of In-In bonds at the surface of InP layers (e.g. a phosphorus depletion occurred at the InP surface). On the other hand, He plasmas appear to be less invasive for the surface. We also noticed that, if concentrated HCl solutions (e.g. > 4 M) were efficient for reducing InP oxides, they lead to a significant modification of the surface morphology and the appearance of boat-like shape of the etch pits. Reducing the HCl concentration allows removing the P and In oxides without damaging the surface. Finally, wet chemical cleaning followed by in situ treatments showed that He direct plasma exhibits the best compromise in terms of surface impact and oxide removal.

We have investigated the impact of various plasma treatments on InGaAs and InP layers. Wet chemical cleaning in concentrated HCl solutions followed by argon or helium direct plasmas is efficient to remove InGaAs native oxides. InP surfaces are more sensitive to plasma treatment than the InGaAs ones but He direct plasma preceding by diluted HCl solution cleaning appears to offer a good compromise between surface deterioration and oxide removal.