New regularization algorithm for DLTS data analysis in amorphous and polycrystalline insulators and semiconductors

Correlation between bismuth concentration and distribution of relaxators in As2Se3(Bi)x layers

“…The obtained values of H(s) were used in the calculation of localized states density N(E) in terms of the method proposed in [8]. As seen from the obtained curve of states density N(E) the spectrum of localized states has changed due to modifications in the glass structure at the value x = 5 at.% Bi (Fig.…”

Oxygen lone-pair states near the valence band edge of aluminum oxide thin films

“…For the as-grown film, the DOS distribution, N(E), could be deconvolved from the original DLTS data according to special ''regularization'' algorithm given in Ref. 9. As shown in Fig.…”

On Distributions of Defect States in Low-k Carbon Doped Silicon Dioxide Films in Vicinity of Fermi Level

“…1 Thus, experimental investigation of electronic spectrum of defect states becomes equally important for physics and applications of low-k insulating films. Deconvolution of the N(E) function from experimentally obtained data requires solution of a one-dimensional integral equation ͑IE͒ of the first kind: e.g., Fredholm's IE appears at the N(E) derivation from field-effect data, 2 Volterra IE links the N(E) curve and spectral dependence of a signal, measured by constant photocurrent method and photothermal deflection spectroscopy, 3 transient photocurrent technique leads to a combination of the Fredholm and Volterra IEs, 4 whereas Mellin IE has to be resolved at the N(E) deconvolution from deep-leveltransient spectroscopy ͑DLTS͒ data, 5 and so on. All such equations are ill-posed in the classical ͑Hadamard͒ sense, 6 when even small ͑of about few per cents͒ errors in the measurements of experimental data can cause huge ͑up to several orders of the magnitude͒ inaccuracy in determination of parameters and shape of the wanted N(E) distribution.…”

“…6 However, quality of such approximation improves with experimental errors reduction, and the regularized N(E) distribution becomes exact at the total absence of such errors. 6,5 Thus, it would be useful to compare the N(E) curves obtained by our regularization method with similar distributions, derived independently ͑e.g., via alternative measurement and deconvolution techniques͒ from the same low-k samples. Such studies are pioneering and important because of very few published papers on experimental investigations of the defect state distribution in low-k insulating materials up to now.…”