Reassessment of the recombination properties of aluminium–oxygen complexes in n‐ and p‐type Czochralski‐grown silicon

Abstract

Phone: þ61 2 612 555 38, Fax: þ61 2 6125 0506The recombination parameters of aluminium-oxygen complexes in silicon have been reassessed by applying lifetime spectroscopy on several n-and p-type intentionally Al-contaminated and control samples, using a single-level defect model. The presence of the control samples has allowed greater accuracy for the extraction of the recombination lifetime. The uncertainty ranges of the parameters have been tightened significantly by simultaneously fitting the lifetime on several samples. The electron/hole capture cross section ratio k was reassessed to be 380, in the uncertainty range of 330-460. A direct comparison of the n-and p-type samples has shown that those complexes are much more recombination-active in p-type silicon than in n-type silicon at low and intermediate injection levels.

Defect Parameters Contour Mapping: A Powerful Tool for Lifetime Spectroscopy Data Analysis

“…This assumption made a priori is thus fundamental to their analysis. On the contrary, the DPCM method is capable of providing the same results by analyzing the τ Al‐O data alone (see Figure of Sun et al) with no assumptions needed. Figure shows the DPCM graph obtained from the modeling of τ Al‐O data along with the metal defects’ parameters taken from Macdonald and Geerligs, Diez et al, Sun et al, Graff, Wang and Sah, Fazzio et al, Roth et al, Kwon et al, and Mishra …”

“…We first apply the DPCM method to a set of IDLS data which represents the simplest possible experimental scenario as the T ‐dependence of the physical parameters evaluated is not taken into account. In particular, we will consider the work of Sun et al in which the authors present a study of the aluminum‐oxygen (Al‐O) recombination center parameters in n ‐ and p ‐type Czochralski–grown silicon. The aluminum‐oxygen center has been extensively studied via deep level transient spectroscopy (DLTS) in the past and E t has been assessed to lay in the range of E t = E v + (0.38 − 0.50) eV whereas k values reported in literature span over orders of magnitudes .…”

“…IDLS‐DPCM analysis of τ Al‐O data reported by Sun et al The Al‐O defect is placed at E t = E v + 0.44 eV and k = 380. The error bars associated to E t and k represent the range of values reported in literature from DLTS analysis and the uncertainty evaluated by Sun et al, respectively. The area contained in between the yellow dashed lines represents the portion of the parameters space where a best fit for τ eff data is found.…”

“…The resulting IDLS‐DPCM plot is characterized by a single bright band with ARV < 5% in the high k ‐region of the parameters space. The plot shows the Al‐O defect placed at E t = E v + 0.44 eV and k = 380 with error bars representing the range of values assessed from DLTS analysis for the former, and the uncertainty reported in Sun et al for the latter. Remarkably, despite no assumptions were made on E t , the DPCM method results match with those presented by Sun et al with an optimal k value of 380 and an equal level of accuracy associated with it, as demonstrated by the width of the best fit region (brightest area) overlapping to their estimated range of uncertainty.…”

“…Lifetime spectroscopy usually refers to injection‐dependent lifetime spectroscopy, i.e., IDLS, if performed at room temperature or temperature‐ and injection‐dependent lifetime spectroscopy, i.e., TIDLS, when the measurements are performed across a range of temperatures it proves the most effective in accurately determining the fundamental defect parameters such as its energy level in the band gap E t , the capture cross section ratio k ( σ n / σ p ), and the T ‐dependence of σ p and σ n . However, most of the reported studies make use of intentionally contaminated samples which means that, although extremely important to expand the general understanding about metal impurities impact on Si material, they rely on the a priori knowledge of the impurity identity introduced into the samples, and oftentimes its concentration.…”