Kruget al.Reply:

Abstract

Krug et al. Reply: In a Comment [1] on our recent Letter [2], Copel discusses features in our XPS data that had not been explained, presents new XPS and MEIS data, and questions the validity of the diffusion-reaction model that we proposed.The first point raised by Copel is the origin of the Si 2p photoelectron peak at 98.5 eV binding energy appearing in the analysis of Al 2 O 3 films on Si(001) after annealing in O 2 at 800 ± C for 30 s, which was unclear in our Letter [2]. Thickness inhomogeneities [1] should indeed be the reason for the above-mentioned peak, which is absent from Copel's results as well as from our results for annealing at lower temperatures. Film inhomogeneities could be induced by extrinsic impurities, as suggested by Copel, or avoided by annealing in ultrahigh vacuum prior to oxidation, as performed by Copel [1]. Such annealings [3] densify the films, reduce oxygen diffusivity during subsequent oxidation, and may result in the observed enhanced stability. In this case, it would be Copel's merit to have found optimal preoxidation annealing conditions.Given the presence of film inhomogeneities in the sample oxidized at 800 ± C in our work, the assumptions underlying our diffusion-reaction model are questioned [1]. On the contrary, we understand that by detecting Si migration and fixation in the Al 2 O 3 film in essentially thickness inhomogeneity-free conditions, Copel provided an important confirmation of our model adequacy. In fact, without oxygen diffusion and reaction one cannot explain Si migration, partial replacement of Al by Si, and Al loss (not entirely quantified in [1]). Furthermore, the specific criticism addressed in the last sentence of the Comment [1] concerning oxygen removal from Al 2 O 3 by excess Si can also be justified. An estimate of the coefficients in the model equations from experimental data is required in order to quantify the relative contributions of different diffusion and reaction processes. Since this information is not yet available, we used the simplest assumption of reactions yielding stoichiometric compounds according to thermodynamic stability data, which is not necessarily valid. Unfortunately, due to partial superposition of Al and Si signals, the MEIS technique used by Copel did not add the necessary information in this sense.An additional, important issue [4] not addressed by us or by Copel regards oxygen diffusion from the gas phase to the Al 2 O 3 -SiO 2 ͞Si interface. This constitutes a basic assumption of the model, as oxygen is taken to trigger Si diffusion and subsequent reaction in the Al 2 O 3 network. In order to investigate oxygen transport, we annealed Al 2 O 3 films on Si(001) in 18 O 2 for various times at 700 ± C, a temperature at which we observed Si migration and fixation without any effect attributable to thickness inhomogeneities. Much thicker (36 nm) films were deposited and oxidized as before [2], aiming at privileging information from the surface and bulk of the films, without contribution of solid-solid interfaces. Excitation curve...

Oxygen reaction-diffusion in metalorganic chemical vapor deposition HfO2 films annealed in O2

“…O 2 annealing for 60 s ͑sample 3 in Table I͒ produces an increase on the O-areal density of about 30% with respect to the as-prepared sample ͑sample 1͒ and a comparable percentage decrease on the N-areal density. Figure 1͑b͒ shows Si, 16 [9][10][11][12][13] in aluminum, zirconium, and gadolinium oxides and silicates, the present HfO 2 /SiO x N y structure displays higher resistance to oxygen migration from the gas into the solid phase and incorporation therein, as well as smaller isotopic exchanges. 29 Si profiles were determined by NRP 16 ͑results not shown here͒, revealing that Si remains immobile during annealing, in contrast to several of the earlier mentioned materials where substrate Si is seen to migrate into the oxide film.…”

“…Previous studies 1 in other high-k ultrathin films on Si indicated that the interface layer thickness and composition can vary according to the deposition method and routine, and various species can be transported during post deposition thermal processing, like oxygen, 9,10 Si, [11][12][13] and the metal species, 14,15 altering the atomic concentrations as well as chemical composition of the system and consequently electrical characteristics like dielectric constant, interface density of states, and mobility of charge carriers in the transistor channel. In particular, stability against annealing in O 2 -containing atmospheres is of high interest, since in further processing steps it is either performed intentionally to improve leakage current and CET characteristics 1,7,8 or unintentionally, because oxygen is almost always residual in any production furnace.…”

Electron spin resonance observation of Si dangling-bond-type defects at the interface of (100) Si with ultrathin layers of SiOx, Al2O3and ZrO2

“…The research enjoys accelerating interest [3,4], with an immense arsenal of complementary high sensitivity analysing techniques being applied in conjunction to explore microstructural, compositional, and bonding chemistry aspects of ultrathin (sub-5 nm) high-κ films on Si [5][6][7][8]. Applied techniques include medium-energy ion scattering (MEIS), high-resolution transmission electron microscopy (HRTEM), nuclear resonance profiling (NRP) and x-ray photoelectron and Auger spectroscopy.…”

Magnetic coupling in highly ordered NiO/Fe 3 O 4 (110): Ultrasharp magnetic interfaces vs. long-range magnetoelastic interactions

“…Therefore, one would in fact expect collinear coupling for arbitrary interface orientations. In another study, however, we found spinflop coupling for Fe 3 O 4 (001)/NiO [24]. In general, the (001)-interface is atomically compensated and spin-flop coupling should therefore occur quite naturally, but the lifting of the NiO sublattice degeneracy by the inverse spinel structure of the magnetite discussed above should suppress this type of coupling.…”