Post Deposition Annealing Atmosphere Effect on SSI-LEDs Made of Zr-Doped HfO2 High-k Gate Dielectric

Monday, October 12, 2015: 12:15
105-C (Phoenix Convention Center)
S. Zhang (Texas A&M University), Y. Kuo (Texas A&M University), and S. Verkhoturov (Texas A & M University)
Recently, Kuo’s group proposed a new type of solid state incandescent light emitting device (SSI-LED) that has a simple MOS structure and emits the warm white light. This kind of device contains a high-k dielectric gate dielectric thin film, such as Zr-doped HfO2 (ZrHfO) or the HfOx [1-3]. The light emission is due to the thermal excitation of many nano-size conductive paths formed from the dielectric breakdown process when the applied gate voltage (Vg) is above a breakdown voltage (VBD). The post deposition annealing (PDA) step is critical to the MOS device performance because it affects the defect structures and densities in the high-k stack [4]. In this study, authors investigated the electrical and optical characteristics of the SSI-LED made of ZrHfO high-k films under different PDA atmospheres.

The ZrHfO high-k films were sputter deposited on the p-type Si (100) (1015 cm-3) wafer using the one pump down process without breakdown the vacuum. The ZrHfO film was deposited from a ZrHf (12/88 wt. %) target in Ar/O2 (1:1) at 5 mTorr and 60 W for 12 min. After deposition, the sample was annealed at 800 °C in N2 or O2 for 3 min. An ITO layer was deposited on the high-k stack and etched into 300 µm dots as the gate electrodes. The backside of the wafer was sputter deposited with an aluminum layer for the ohmic contact purpose. The complete sample was annealed at 400 °C in forming gas for 5 min. The complete sample was characterized for electrical and optical properties.

Figure 1 shows the I-V curves of the N2 PDA and O2 PDA devices with Vg swept from 0 to -10 V. The devices break abruptly at -5.65 V for the N2 PDA sample and -6.1 V for the O2 PDA sample. The O2 PDA sample has a larger physical thickness than that of the N2 PDA sample, e.g., the equivalent oxide thickness (EOT) of the former is 12.03 nm and that of the latter is 11.12 nm, which can explain the difference of the VBD’s. In addition, the breakdown strength of the ZrHfO stack is related to the defect densities in the bulk high-k film and the Si/high-k interface layer [5]. The oxide trapping density (Qot) and the interface density of states (Dit) of the O2 PDA sample are 5.21×109 cm-2 and 1.61×1011 cm-2∙eV-1, separately, which are smaller than those of the N2 PDA sample, i.e., 7.13×109 cm-2  and 1.80×1011 cm-2∙eV-1, separately, which also favors of high VBD of the former sample.   

Figure 2 shows the high magnification light emission photos of the N2 PDA and O2 PDA devices stressed at Vg = -50 V. Both photos show that the light is emitted from many discrete bright dots uniformly distributed across the gate electrode. The N2 PDA sample contains a larger number of weaker dots than the O2 PDA sample does. Since the bright dots are from the thermal excitation of conductive paths in the sample, the number density of the bright dots is related to the breakdown strength of the sample. The O2 PDA sample has lower defect densities in the bulk ZrHfO and HfSiOx interface layers than the N2 PDA sample [5]. In addition, bright dots in Fig. 2(b) appear to be larger than most of those in Fig. 2(a). The bright dots at the edge of the sample appear to be larger than those in the center. The difference in the size distribution of the bright dots is related to the breakdown strength the high-k stack.       

Figure 3 shows the emission spectra of the N2 PDA and O2 PDA samples stressed at Vg = -50 V. Both spectra cover the same wavelength range including the visible and some of the near IR lights. The O2 PDA sample has a higher light intensity than the O2 PDA sample has, which is consistent with the result of Fig. 2. The peak wavelengths of both samples are very close, i.e., 681.5 nm for the O2 PDA sample and 682.0 nm for the N2 PDA sample. The similar composition of the conductive paths in both samples may be the cause of the spectrum shape.

[1]   Y. Kuo and C.-C. Lin, APL, 102, 031117 (2013).

[2]   Y. Kuo and C.-C. Lin, ESSL, 2, Q59 (2013).

[3]   Y. Kuo and C.-C. Lin, SSE, 89, 120 (2013).

[4]   G. D. Wilk, R. M. Wallace, and J. M. Anthony, JAP, 87, 484 (2000).

[5]   C.-C. Lin and Y. Kuo, EJSSST, 3, Q182 (2014).