No abstract

An energy level of i^ + 0.38 eV related to the 0 to + charge transition of" substitutional manganese in silicon has been determined with a combination of deep-level-transient spectroscopy and ESR measurements. There is no evidence for ordinary amphoteric or negative-U behavior in the lower half of the band gap. This is the first identified energy level of a 3 d substitutional impurity in silicon.PACS numbers: 71.55.FrThere is a long tradition of exploring the properties of 3d (iron group) transition-metal point defects in silicon (see, for example, Weber 1 and references therein). Earlier calculations of the charge-transition levels including spin polarization and electronic relaxation were applied only to interstitial 3d elements. 2 Recently, advanced self-consistent calculations using a spin-unrestricted linear-muffin-tin Green's function method were carried out and expanded to substitutional 3delements. 3,4 Dependable data concerning the energy levels of the interstitial site could be determined by use of deep-level transient spectroscopy (DLTS) in combination with Hall and ESR measurements. 5 " 8 Regarding the substitutional 3d elements, however, no well-identified data were available so far. Thus the value of the 0 to + charge-transition level of substitutional manganese presented in this paper can serve as a first check of the validity of existing theories when applied to substitutional 3 d transition metals, and might give valuable hints for further investigations.To form substitutional manganese, samples were made from 1.7-ft cm /?-type silicon by application of a copper-manganese codiffusion technique as reported by Ludwig and Woodbury. 9 Successfully prepared samples show the typical ESR fingerprint [manganese nuclear spin (MNS)] of single positively charged substitutional manganese. 9 The X-band spin-resonance first-derivative absorption signal shown in Fig. 1 exhibits full cubic symmetry. The twofold spin degeneracy (5=1) is lifted by interaction with the manganesenuclear-spin system (/=- §-), leading to a 2x6 line splitting. The Hamilton parameters (# = 2.0259, A = -63.09x 10~4 cm -1 ) are different from those of a signal, possessing the same manifold and symmetry, which is attributed to negatively charged interstitial manganese. 9 All samples containing substitutional manganese produce both the MNS signal and a DLTS 10 peak exhibiting an energy level of 0.38 ±0.01 eV as derived from the Arrhenius plot of the DLTS spectra measured at different lock-in frequencies. Reference samples prepared with copper diffusion but without manganese codiffusion show no detectable 0.38-eV peak. On the other hand, samples could be produced that show the MNS signal and contain only the 0.38-eV transient with no other visible residual peaks. A correlation plot (Fig. 2) shows the unambiguous proportionality between the ESR signal intensity, obtained by double integration of the MNS signal, and the corresponding calibrated DLTS peak height. Each point is gained from an individual sample with a different concentration of s...

No abstract

A combination of deep level transient spectroscopy (DLTS) and electron spin resonance (ESR) measurements was used to determine the energy level of substitutional manganese in silicon. Samples of p-type silicon were subjected to a copper-manganese codiffusion. Successfully prepared samples show the typical ESR signal of substitutional manganese with a single positive charge. The Hamiltonian parameters g=2.029 and A=−62.7×10−4 cm−1 are different from those for negatively charged interstitial manganese. The DLTS measurements reveal an energy level of M1=0.39 eV above the valence-band edge for the substitutional manganese. Because of the codiffusion of Cu also the previously reported levels C1=0.098 eV, C2=0.22 eV, and C3=0.41 eV were found. The combination of ESR and DLTS results allowed a conclusive identification of the defect level M1 and provided no evidence for ordinary amphoteric or negative U behavior in the lower half of the band gap. Furthermore, isothermal and isochronal annealing experiments were performed which support the conclusion that the defect level M1 orginates from substitutional manganese in p-type silicon.

No abstract