Germanium MOS Capacitors with Hafnium Dioxide and Silicon Dioxide Dielectrics

Germanium (Ge) does not grow a suitable oxide for MOS devices. The Ge/dielectric interface is of prime importance to the operation of photo-detectors and scaled MOSTs. Therefore there is a requirement for deposited or bonded dielectric materials. MOS capacitors have been formed on germanium substrates with three different dielectric materials. Firstly, a thermally grown and bonded silicon dioxide (SiO2) layer, secondly, SiO2 deposited by atmospheric pressure CVD ‘silox’, and thirdly a hafnium oxide (HfO2) high-k dielectric deposited by atomic layer deposition (ALD). Ge wafers used were p-type 1 0 0 2 O cm. C–V measurements have been made on all three types of capacitors to assess the interface quality. ALD HfO2 and silox both display acceptable C–V characteristics. Threshold voltage and maximum and minimum capacitance values closely match expected values found through calculation. However, the bonded SiO2 has non-ideal C–V characteristics, revealing the presence of a high density of interface states. A H2/N2 post metal anneal has a detrimental effect on C–V characteristics of HfO2 and silox dielectrics, causing a shift in the threshold voltage and rise in the minimum capacitance value. In the case of hafnium dioxide, capacitor properties can be improved by performing a plasma nitridation of the Ge surface prior to dielectric deposition.

Conformal oxide coating of carbon nanotubes

The International Roadmap for Ferroelectric Memories requires three-dimensional integration of high-dielectric materials onto metal interconnects or bottom electrodes by 2010. Here, we demonstrate the possibility of conformally coating carbon nanotubes with high-dielectric oxide as a first step toward ultrahigh integration density of three-dimensional ferroelectric random access memories.

Intrinsic dielectric response in ferroelectric nano-capacitors

Measurements on 'free-standing' single-crystal barium titanate capacitors with thickness down to 75 nm show a dielectric response typical of large single crystals, rather than conventional thin films. There is a notable absence of any broadening or temperature shift of the dielectric peak or loss tangent. Peak dielectric constants of similar to25 000 are observed, and Curie-Weiss analysis demonstrates first order transformation behaviour. This is in dramatic contrast to results on conventionally deposited thin film capacitor heterostructures, which show large dielectric peak broadening and temperature shifts (e.g. Parker et al 2002 Appl. Phys. Lett. 81 340), as well as an apparent change in the nature-of the paraelectric-ferroelectric transition from first to second order. Our data are compatible with a recent model by Bratkovsky and Levanyuk (2004 Preprint cond-mat/0402100), which attributes dielectric peak broadening to gradient terms that will exist in any thin film capacitor heterostructure. The observed recovery of first order transformation behaviour is consistent with the absence of significant substrate clamping in our experiment, as modelled by Pertsev et al (1998,Phys. Rev. Lett. 80 1988), and illustrates that the second order behaviour seen in conventionally deposited thin films cannot be attributed to the effects of reduced dimensionality in the system, nor to the influence of an intrinsic universal interfacial capacitance associated with the electrode- ferroelectric interface.

The effect of flexoelectricity on the dielectric properties of inhomogeneously strained ferroelectric thin films

Recent experimental measurements of large flexoelectric coefficients in ferroelectric ceramics suggest that strain gradients can affect the polarization and permittivity behaviour of inhomogeneously strained ferroelectrics. Here we present a phenomenological model of the effect of flexoelectricity on the dielectric constant, polarization, Curie temperature (T-C), temperature of maximum dielectric constant (T-m) and temperature of the onset of reversible polarization (T-ferro) for ferroelectric thin films subject to substrate-induced epitaxial strains that are allowed to relax with thickness, and the qualitative and quantitative predictions of the model are compared with experimental results for (Ba0.5Sr0.5)TiO3 thin films on SrRuO3 electrodes. It is shown that flexoelectricity can play an important role in decreasing the maximum dielectric constant of ferroelectric thin films under inhomogeneous in-plane strain, regardless of the sign of the strain gradient.

Maximum of dielectric permittivity caused by structural transition in ferroelectric BaTiO3-SrTiO3 superlattices

We invoke the onset of dislocations along the BaTiO3-SrTiO3 interface as reported by Wunderlich et al. to explain the non-monotonic behaviour of the dielectric permittivity as a function of superlattice periodicity and the less than four-fold in-plane symmetry at the dielectric maximum. At a periodicity of about 10/10, depending on composition and growth mechanism, several groups report a maximum of dielectric permittivity. In addition to that we observe in-plane symmetry less than tetragonal for 10/10 superlattices by HR-XRD, in contrast to initial low-resolution data from Tabata et al. thus challenging the assumption of unrelaxed strain all the way through the superlattice. The aim of this article is to link both effects to the increasing volume fraction of conducting layers close to the interface in series with the superlattice layers.

The mid-infrared dielectric function of NBCO in the a-b plane from 85 K to 300 K

Results are reported on the a-b plane dielectric function (epsilon) of thin-film c-axis NdBa2Cu3O7-delta with close to optimal oxygen doping (T-c similar to 90 K) in the mid-infrared (wavelength 3.392 mum) over the temperature range 85 K to 300 K. An attenuated total reflectance technique based on the excitation of surface plasmon polaritons is used. The results show that \epsilon (r)\ decreases quasi-linearly with increasing temperature, while Ei is invariant with temperature to within experimental uncertainties. Representative values are epsilon = [epsilon (r) + i epsilon (i)] = (-12.9 +/- 0.6) + i(23.0 +/- 1.5) at T similar to 295 K and epsilon = (-15.7 +/- 0.7) + i(23.5 +/- 1.1) at T similar to 90 K. The raw data an interpreted in terms of the generalized Drude model which gives effective scattering rates (1/tau*) that increase with temperature from about 3800 cm(-1) at 90 K to about 4300 cm(-1) at 295 K. There are indications of a superlinear T-dependence in the scattering, 1/tau*: a fit to a function of the form 1/tau* = A + BTalpha gives alpha = 2.8 +/- 0.7. The effective plasma frequency, omega (p)*, with an average value of approximately 21 000 cm(-1) was independent of temperature.