(Invited) Spectroscopy of Deep Gap States in High-k Insulators

Degradation phenomena in metal-insulator-semiconductor devices are often associated with unwanted charging of the insulating layers, prompting efforts to understand the nature of the charge traps. While a large number of theoretical studies regarding energy levels of different defects and impurities can be found in the literature, experiments usually provide only the density and type of the gap states which makes comparison difficult. Here we present a methodology of energy distribution measurements using the Exhaustive Photo-Depopulation Spectroscopy (EPDS) and will demonstrate its successful application to characterization of electron traps in various insulating materials.

The conventional approach to the observation photoioni­zation (PI) of gap states using measurements of photocurrent provides only limited information since the current is determined by the convolution of the energy-dependent trap distribution and the PI cross-section. One can only find the energy onset of the PI but neither the density nor energy distribution of traps can be recovered from the photo-depopulation current spectra. To resolve this problem, the EPDS employs measurements of the insulator charge allowing for the PI to exhaust all charge carriers available for optical excitation at a given photon energy hν. If starting from a low photon energy hν and then increase it by the energy step dhν, the saturation of the de-trapping kinetics at each photon energy range {hν; hν+dhν} would mean that are no electrons left available for optical transitions to the conduction band (CB) in this energy window. Then, the amount of charge de-trapped during the next step exactly corresponds to the density of occupied electron states in the energy interval dhν. By performing the EPDS at incremental photon energies one can reconstruct the distribution of the electron state density across the bandgap of the insulator.

EPDS can be successfully used to determine the distribution of electron states associated with defects and impurities in the bandgap of high-k insulators. The studied samples were prepared on thermally oxidized (100)Si substrates with SiO₂ layers of 4-8.5 nm thickness serving as the tunneling barrier. On top of SiO₂ a 12-20 nm thick layer of another insulator was deposited (Si₃N₄, Al₂O₃, HfO₂, Hf_0.8Al_0.2O_x, LaAlO₃, Gd_xAl_2-xO₃, Y₂O₃) followed by evaporation of semi-transparent metal electrodes (Au, Al, TiN). These samples were used in the EPDS experiments performed starting from the initial value hν=1.3 eV with stepwise increments of dhν=0.3 eV. Each of the samples was analyzed twice: First, in the uncharged (pristine) state to reveal the possible removal of electrons from the initially neutral donor centers, and then, for the second time, after tunnel injection of electrons from the Si substrate through the SiO₂barrier. The charge variation after each illumination step was converted to the spectral charge density (SCD) reflecting the energies of all optically-induced charging and discharging processes in the MIS structure.

Only a marginal amount of positive charge is generated in the uncharged samples suggesting a negligible density of donor states in the studied range of electron levels (<4 eV below the insulator CB) in all the studied insulators. At the same time, electron de-trapping strongly correlates with the density of electrons trapped after tunneling injection suggesting acceptor-type traps to be dominant. The energy distribution of the trapped electron levels is in the 2<E_t<3 eV interval in Hf_0.8Al_0.2O_x, 3<E_t<4 eV in HfO₂, and 2<E_t<4 eV in Al₂O₃ with a particularly high trap density found in the crystallized layers. Somewhat deeper traps (E_t>3 eV) are found in Si₃N₄. As the striking common feature of the studied insulators, the energy distributions of electron levels appear to be at least 1-eV wide suggesting a large impact of structural disorder of the insulating network.Degradation phenomena in metal-insulator-semiconductor devices are often associated with unwanted charging of the insulating layers, prompting efforts to understand the nature of the charge traps. While a large number of theoretical studies regarding energy levels of different defects and impurities can be found in the literature, experiments usually provide only the density and type of the gap states which makes comparison difficult. Here we present a methodology of energy distribution measurements using the Exhaustive Photo-Depopulation Spectroscopy (EPDS) and will demonstrate its successful application to characterization of electron traps in various insulating materials.

The conventional approach to the observation photoioni­zation (PI) of gap states using measurements of photocurrent provides only limited information since the current is determined by the convolution of the energy-dependent trap distribution and the PI cross-section. One can only find the energy onset of the PI but neither the density nor energy distribution of traps can be recovered from the photo-depopulation current spectra. To resolve this problem, the EPDS employs measurements of the insulator charge allowing for the PI to exhaust all charge carriers available for optical excitation at a given photon energy hn. If starting from a low photon energy hν and then increase it by the energy step dhn, the saturation of the de-trapping kinetics at each photon energy range {hν; hν+dhν} would mean that are no electrons left available for optical transitions to the conduction band (CB) in this energy window. Then, the amount of charge de-trapped during the next step exactly corresponds to the density of occupied electron states in the energy interval dhn. By performing the EPDS at incremental photon energies one can reconstruct the distribution of the electron state density across the bandgap of the insulator.

EPDS can be successfully used to determine the distribution of electron states associated with defects and impurities in the bandgap of high-k insulators. The studied samples were prepared on thermally oxidized (100)Si substrates with SiO₂ layers of 4-8.5 nm thickness serving as the tunneling barrier. On top of SiO₂ a 12-20 nm thick layer of another insulator was deposited (Si₃N₄, Al₂O₃, HfO₂, Hf_0.8Al_0.2O_x, LaAlO₃, Gd_xAl_2-xO₃, Y₂O₃) followed by evaporation of semi-transparent metal electrodes (Au, Al, TiN). These samples were used in the EPDS experiments performed starting from the initial value hν=1.3 eV with stepwise increments of dhν=0.3 eV. Each of the samples was analyzed twice: First, in the uncharged (pristine) state to reveal the possible removal of electrons from the initially neutral donor centers, and then, for the second time, after tunnel injection of electrons from the Si substrate through the SiO₂ barrier. The charge variation after each illumination step was converted to the spectral charge density (SCD) reflecting the energies of all optically-induced charging and discharging processes in the MIS structure.

Only a marginal amount of positive charge is generated in the uncharged samples suggesting a negligible density of donor states in the studied range of electron levels (<4 eV below the insulator CB) in all the studied insulators. At the same time, electron de-trapping strongly correlates with the density of electrons trapped after tunneling injection suggesting acceptor-type traps to be dominant. The energy distribution of the trapped electron levels is in the 2<E_t<3 eV interval in Hf_0.8Al_0.2O_x, 3<E_t<4 eV in HfO₂, and 2<E_t<4 eV in Al₂O₃ with particularly high trap density found in the crystallized layers. Somewhat deeper traps (E_t>3 eV) are found in Si₃N₄. As the striking common feature of the studied insulators, the energy distributions of electron levels appear to be at least 1-eV wide suggesting a large impact of structural disorder of the insulating network.