Appl. Phys. Lett., Vol. 73, No. 20, 16 November 1998
Zhuravlev, Gilinsky, and Kobitsky
2963
diative channels. In the opposite case of excitonic recombi-
nation in quantum dots, the saturation of the ground-level PL
is accompanied by the onset of the PL of the excited states of
the dots.9,10 Since we do not observe any change in the spec-
tral shape of the band with excitation intensity, a sublinear
dependence should not be expected for the recombination
between quantum-confined levels in nanocrystals in our ex-
perimental conditions. Third, the absence of the spectral de-
pendence of PL kinetics also points to recombination via
local centers. On the contrary, in the case of the quantum-
confinement recombination model, the PL decay should pro-
ceed faster on the shorter wavelength side of the spectrum
because of the strong dependencies of both the recombina-
tion probability and energy of optical transitions on nano-
crystal radii,7,11 resulting in a redshift of the band with time.
Let us speculate on the nature of the recombination cen-
ter. Recently, a center of recombination on the Si
nanocrystal–silicon dioxide boundary that is responsible for
1.5 eV luminescence was considered by Allan et al.6 They
have shown that a single covalent bond, for example a Si–Si
bond, can act as a trap and a recombination center. Their
calculation shows that a metastable recombination state,
separated from the excited state by an energy barrier, can
exist on the boundary of small nanocrystals. The nonexpo-
nential decay kinetics can be expected in such centers if the
barriers between excited and metastable states of the centers
would have a certain energy spread and, consequently, the
probability of carrier transition from the excited to recombi-
nation state would differ. In our case, since the recombina-
tion centers are localized on the boundaries of nanocrystals,
the barrier height can depend on the local environment.
A particular qualitative characteristic of the center can
be inferred from the temperature dependence of its PL band
energy. It is known that the temperature dependence of the
band energy of a center that is strongly bound with crystal
lattice differs from that of the band gap of the host crystal.
According to the configuration diagram model,12,13 the direc-
tion of the PL band shift with temperature is determined by
the ratio of the frequencies of vibration modes of the ground
and excited states of the center. The blueshift of the band
with decreased temperature that we observe experimentally
enables us to conclude that in our case the frequency of the
vibration mode of the ground state of the center is higher.
In conclusion, we have studied the steady-state and time-
resolved luminescence of 3.5 nm silicon nanocrystals fabri-
cated by Si ion implantation into a SiO2 matrix. We observed
that the 1.5 eV PL band that is usually attributed to the
recombination of quantum-confined carriers in silicon quan-
tum dots displayed a sublinear intensity dependence on the
excitation power, a spectrally uniform decay after transient
excitation, and besides that, the estimated energy position of
the band differs strongly from the experimentally observed.
These data enable us to conclude that the characteristic 1.5
eV luminescence of Si nanoclusters is governed by the re-
combination via the levels of some defect-related centers,
which are presumably localized at the silicon nanocrystal–
silicon dioxide boundary.
FIG. 2. Dependence of the steady-state PL intensity of the nanocrystals on
the excitation power at room temperature.
In Fig. 3 a set of time-resolved PL spectra measured at
300 K is shown as a function of the delay time after the
excitation pulse. It is seen that the shape of the spectrum
does not change with the delay time. The inset to Fig. 3
shows decay curves integrated over the spectrum taken at
several temperatures on a wider time scale. At all tempera-
tures the decay is nonexponential, the curves are approxi-
mated by stretched exponential functions. The 1/e decay
time is 450, 65, and 20 s at 4.2, 77, and 295 K, respec-
tively. Though the temperature decrease leads to a consider-
able increase of the decay time accompanied by a slight in-
crease of the steady-state PL intensity, we did not observe
any change of the band shape with delay time in this tem-
perature.
The above results support the assumption that radiative
recombination in silicon nanocrystals is mediated by some
defect levels. The following observations point to this con-
clusion. First, the energy of radiative recombination is con-
siderably less than the expected energy of optical transitions
between the quantum-confined levels ͑see Fig. 1͒. Second,
the sublinear dependence of PL intensity on the excitation
power indicates that the recombination is mediated by some
localized centers which saturate at high excitation powers8
and thus let charge carriers recombine via competing nonra-
FIG. 3. Evolution of the transient PL spectrum of Si nanocrystals with time
at room temperature. The spectra were taken at delays of 0.25, 1, 3, 8, 15,
25, and 50 s. The inset shows the decay curves integrated over the band
The authors are grateful to Dr. W. Skorupa, Dr. G. A.
Kachurin, and Dr. I. E. Tyschenko for supplying them with
the samples used in this study and to Dr. A. K. Gutakovsky
spectrum at 4.2, 77, and 300 K ͑from upper to lower͒.
132.174.255.116 On: Wed, 26 Nov 2014 19:05:37