Elsevier

Optical Materials

Volume 27, Issue 5, February 2005, Pages 884-889
Optical Materials

Time dependence and excitation spectra of the photoluminescence emission at 1.54 μm in Si-nanocluster and Er co-doped silica

https://doi.org/10.1016/j.optmat.2004.08.029Get rights and content

Abstract

The risetime of the 1.54 μm luminescence emission in Er:Si-nc co-doped glasses provides information on the mechanisms leading to population of the luminescent level, i.e. on the energy transfer time. In this paper we present a detailed study of this risetime in silica glasses after excitation in the wavelength region 360–860 nm by a continuously tunable pulsed Ti:Al2O3 laser. The emission risetime turns out to be dependent on the sample fabrication method used, i.e. implantation or co-sputtering and on the annealing temperature used to precipitate the silicon aggregates. The fastest energy transfer time observed in our samples is τtr  2 μs. The risetime was found to be weakly dependent on the excitation wavelength up to around 700 nm where a sudden increase is observed. This increase is more pronounced in samples where the annealing temperature was such to produce amorphous aggregates rather than crystalline ones. The difference in the nature of electronic states excited by near-infrared pumping in the two types of samples containing crystalline or amorphous aggregates is also evidenced by CW photoluminescence excitation (PLE) measurements which reflect the absorption coefficient of the sensitizing centers. In samples containing amorphous aggregates we observe an exponential dependence of the PL emission intensity versus excitation energy, which is indicative of an Urbach tail due to the disordered structure of aggregates. In the sample containing crystalline aggregates we observe an indirect interband transition behavior, with an energy gap of 1.56 eV.

Introduction

Sensitization of the Er3+ photo-luminescence emission at 1.54 μm through energy transfer from Si quantum dots (Si-nc) has attracted much interest because of potential application in the fabrication of broadly pumpable Er-doped waveguide amplifiers [1], [2], [3], [4]. In fact these systems efficiently emit 1.54 μm radiation when excited in the blue–green part of the spectrum out of resonance with the Er3+ ion levels [5]. Furthermore the sensitization process produces an increase of the effective absorption cross-section of erbium of about three orders of magnitude [6]. The sensitization process does not appear to be only connected to the presence of crystalline silicon nanoparticles, since amorphous aggregates act as sensitizers as well [7].

Despite the assessed importance of the silicon excess in the matrix, investigation on several aspects of the energy transfer mechanism is still at early stages, especially concerning the excitation dynamics. For example the silicon to erbium transfer time, governing the saturation level of the excitation flux [10] and thus the achievable optical gain, and the energy levels involved in the transfer process are still subject of debate [8], [9], [10]. Moreover, a detailed comparison of the sensitizing ability of amorphous and crystalline sensitizers is lacking. This is partly due to the use of excitation sources emitting on discrete wavelengths in most of the published studies.

In this paper we report on the spectroscopic investigation of the energy transfer mechanism in Si-nc/Er co-doped glasses using both the fundamental and the frequency-doubled emission of a continuously tunable nanosecond Ti:Al2O3 laser. The use of a short pulse laser offers the possibility to estimate the energy transfer time from the risetime of the 1.54 μm emission in a co-doped material. In principle this risetime is determined by two components, the transfer time from Si-nc to a certain Er3+ energy level, and the intracenter decay time from the pumped level to the luminescent one, i.e., the 4I13/2. The second component, i.e., the intracenter decay time, can in turn be estimated from the risetime of the 1.54 μm emission in a singly-doped material when excited on an Er3+ level. Thus, from comparison of the 1.54 μm luminescence risetimes in singly-doped and co-doped materials we can get an estimate of the Si-nc to Er3+ transfer time.

Samples used in this study were prepared by two different techniques: erbium and silicon co-implantation in bulk silica or magnetron co-sputtering. Two co-sputtered samples annealed at different temperature to precipitate non stoichiometric silica are compared to evidence the effect of silicon morphology on the Er sensitization [11]. To this aim we also performed CW PLE measurements in the visible and NIR regions of the spectrum.

Section snippets

Experimental

Samples used in this study are Er implanted (sample a1) and Si-nc:Er co-implanted (sample a2) silica, and Si-nc:Er co-sputtered silica annealed at two different temperatures, namely 900 °C (sample b1) and 1100 °C (sample b2). Details on the preparation of the implanted sample are reported in Refs. [4], [7]. Briefly, the silica substrate was implanted with 80 keV Si+ to a dose of 1 × 1017/cm2 and 300 keV Er3+ to a dose of 1 × 1015/cm2. The implant energies of the two elements were chosen to optimize the

Results

By exciting the Er implanted sample (sample a1) at 380 nm in resonance with the 2G9/2 level we get a 1.54 μm emission risetime of (3.3 ± 0.3) μs, resulting from the intracenter cascade decay from the excited level to the luminescent one. By using out of resonance excitation we were not able to detect any signal. The same measurement on the Er and Si co-implanted sample (sample a2) yield a risetime of (2.3 ± 0.4) μs, irrespective of the pumping wavelength from 370 to 400 nm. The error is calculated from

Discussion

Due to the peculiar energy level structure of the Er3+ ion, the 1.54 μm emission risetime in the singly-doped system excited in the visible region is mostly due to the 4I11/2 to 4I13/2 multiphonon transition. In fact in cascaded multiphonon transitions the decay time is determined by the step which has the largest energy gap., i.e., the so-called “rate limiting step”, the other steps playing a negligible role [14]. Consequently this risetime must be almost constant for excitation on all the

Conclusions

In conclusion the risetime of the 1.54 μm emission in co-doped samples is faster than the intracenter decay in singly-doped samples for visible excitation. This could be indicative of a very fast energy transfer process and/or of a substantial modification of intracenter decay rates in the co-doped material. Furthermore the 1.54 μm emission risetime turns out to be dependent on annealing temperature and sample preparation method. Measurements of the excitation wavelength dependence of the 1.54 μm

Acknowledgement

M.F. would like to thank A. Mittiga for help and discussion. This work was supported by the EU SINERGIA project (Contract No. IST 2000-29650).

References (27)

  • C. Ternon et al.

    Thin Solid Films

    (2002)
  • Y. Kanemitsu et al.

    Thin Solid Films

    (2001)
  • V. Ligatchev

    Physica B

    (2003)
  • A.J. Kenyon et al.

    J. Phys.: Condens. Matter

    (1994)
  • U. Hommerich et al.

    Appl. Phys. Lett.

    (1997)
  • M. Fujii et al.

    J. Appl. Phys.

    (1998)
  • G. Franzò et al.

    Appl. Phys. A: Mater. Sci. Process.

    (1999)
  • A.J. Kenyon et al.

    Appl. Phys. Lett.

    (2000)
  • F. Priolo et al.

    J. Appl. Phys.

    (2001)
  • G. Franzò et al.

    Appl. Phys. Lett.

    (2003)
  • K. Watanabe et al.

    J. Appl. Phys.

    (2001)
  • M. Fujii et al.

    J. Appl. Phys.

    (2004)
  • D. Pacifici et al.

    Phys. Rev. B

    (2003)
  • Cited by (0)

    View full text