III-V nitride materials combine unique properties such as a direct and tuneable band gap from n-IR (InN, 0.7 eV) to n-UV (AlN, 6.2 eV). Among the electronic devices, AlGaN/GaN and InAlN/GaN device structures are commonly developed for HEMT (high electron mobility transistors). The strong advantage of In_xAl_1-xN is a spontaneous high polarization leading to an extremely high carrier density in 2DEG (about 3.5 x 10¹³ cm^-2). Moreover, the near lattice matched composition with GaN is achieved for x = 0.17 - 0.20, leading to a strain-free heterojunction which drastically reduce the structural defects [1].

As a high difference is encountered in growth temperature between InN (600°C) and AlN (1100°C), the main difficulty resides in making a high crystalline quality and a good homogeneity of InAlN layers [2]. Nevertheless, MOCVD gives excellent composition uniformity and good device performance for ultra thin-layers (from 7 to 9 nm thickness).

III-V nitride materials are known to be chemically stable at working temperature and poorly reactive to air exposure. Nevertheless, higher annealing temperatures (more than 800°C) are needed during engineering when ohmic contact is performed. Such materials request high performances and little variations on the surface chemistry can harm the quality of 2DEG. Moreover, InAlN/GaN structure shows a lower thermal stability compared to AlGaN/GaN devices [3].

Only few studies describe the thermal stability of InAlN layers [4]. Some workers show a better thermal stability for a slower rate of growth in MOCVD [5], whereas others focus on an ohmic contact annealing at a lower temperature of 600°C to 650°C [6].

In this work, we choose to study by XPS the chemical surface engineering on InAlN ultra-thin layers at higher temperature, from 850°C to 950°C, under O₂ rich and Ar atmosphere, in order to detect the possible degradation mechanisms usually running through the HEMT elaboration process.

XPS characterizations lead us to an accurate analyze on the major effects of the thermal treatments. The table Figure 1 (a) shows that the O₂ thermal treatment at 850°C drastically increase the oxygen content and decrease the nitrogen content, justifying the formation of an ultra-thin oxide layer on surface. After the O₂ thermal treatment, carbon content is very low, around 6 at%, showing layers with a very few contamination. The additional Ar thermal treatment at 850°C has practically no impact on the layer.

Concerning the InAlN matrix, an important chemical change is observed at high energy (404 eV) in the N1s band. Figure 1 (b) shows the evolution of the N1s peak for InAlN as-grown layer, and layers annealed 850°C and 950°C under O₂ rich atmosphere for 1 min. Area ratios shows an evolution in the proportion of the additional N1s peak at 404 eV from 32% for the annealing at 850°C to 56% for the annealing at 950°C. The very strong gap of 6.7 eV, observed between the later and the N1s peak at 397.3 eV accredited to InAlN, is attributed to interstitial N₂ molecules trapped in the structure [7]. Enlargement of the N1s band at 397.3 eV can also be seen. This effect could be attributed to N-N bonded defect in structures [8].

The aging of the samples shows that the N₂ interstitial molecules are slowly decreased after 2 months and almost totally gone after 1 year. Unlike this evolution, N-N defects still present inside the matrix even after 1 year of aging justifying that those defects are deeply incorporated inside the InAlN layer.

To our knowledge, this unusual behavior was not reported before on InAlN layers. Indeed, the surface modification using physical process such as N₂ plasma or chemical solution process such as NH₄OH treatment will be also getting into details.

References

[1] J. Kuzmik, IEEE Electron Device Lett., 2001, 22 : 510.

[2] E. Taylor, M.D. Smith, T.C. Sadler, K. Lorenz, H.N. Li, E. Alves, P.J. Parbrook, R.W. Martin, J. Cryst. Growth, 2014, 408 : 97.

[3] H. Yu, M. Ozturk, P. Demirel, H. Cakmak, B. Bolukbas, D. Caliskan, E. Ozbay, Semicond. Sci. Technol., 2011, 26 : 085010.

[4] L. Lugani, J.-F. Carlin, M.A. Py, N. Grandjean, Appl. Phys. Lett., 2014, 105 : 112101.

[5] A. Watanabe, J.J. Freedsman, Y. Urayama, D. Christy, T. Egawa, J. of Appl. Phys., 2015, 118 : 235705.

[6] J. Lee, M. Yan, B. Ofuonye, J. Jang, X. Gao, S. Guo, I. Adesida, Phys. Status Solidi A, 2011, 208 : 1538.

[7] I. Bertóti, Catalysis Today, 2012, 181 : 95.

[8] L. Rosenberger, R. Baird, E. McCullen, G. Auner, G. Shreve, Surf. Interface Anal., 2008, 40 : 1254.