Miniaturizing the erbium-doped optical fiber amplifier (~20 m in length) into a
small, compact amplifier that can be integrated with other optical and
electronic devices on a single chip (optoelectronics) offers great promise in
optical communication as an alternative to the electronic
technology.1,2, The
gain of these miniaturized devices is limited by the solubility, concentration,
and distribution of optically-active Er3+ in a host material.
3 While
incorporation of a high concentration of erbium is possible by ion
implantation, the method does not allow for the control of spatial distribution
or the activation of the ions in the host. This is critical at high Er
concentrations, since non-radiative processes resulting from ion-ion
interactions become dominant and significantly reduce the photoluminescence
(PL) yield.
In this work, we demonstrated radical-enhanced atomic layer deposition (RE-ALD)
as a viable technique to synthesize Er3+-doped dielectric thin films
with a precise control of its concentration and spatial distribution, thus
tailoring the PL property of Er3+ doped waveguides. Since the
optically-active Er needs to be in the trivalent state, showing highest
photoluminescence efficiency when coordinating with approximately six O atoms
as in crystalline Er2O3, Y2O3 was
chosen as the host material due to its identical crystal structure and very
similar lattice constant to Er2O3.4-5 In this case, typical problems such as
lattice distortion and vacancy formation which are detrimental to the PL yield
can essentially be eliminated. The thin film deposition was carried out in an
ultra-high vacuum multi-beam reactor in which metal b-diketonate complexes and oxygen radicals were introduced
independently and sequentially. Incorporation of Er in
Y2O3 thin films at 350°C was accomplished by combining
the self-limiting RE-ALD of Y2O3 and
Er2O3 in an alternating fashion, with the
Er doping level at a specific depth location controlled by varying the ratio of
Y2O3:Er2O3 cycles during deposition.
Figure 1.
(a) Proposed structures of Y2O3 films doped with low and
high concentrations of Er3+, (b) EXAFS analysis of a 6 at.%
Er3+ doped Y2O3
thin film, and (c) EXAFS analysis of a 14 at.% Er3+ doped
Y2O3 thin film.
The nanostructure of Er-doped Y2O3 thin films was
investigated by using a high-resolution transmission electron microscopy
(HRTEM) and electron energy loss spectrometry (EELS). Specifically, the
distribution of Er separated by layers of Y2O3 was
confirmed by elemental EELS mapping of Er M4 and M5,
with the Er concentration controlled from 6 to 14 at.%, determined by X-ray
photoelectron spectroscopy (XPS). This unique
feature is characteristic of the alternating RE-ALD of
Y2O3 and Er2O3. The
photoluminescence yield was found to reduce by at least one order of magnitude
when the Er doping level exceeded 8 at.%. This photoluminescence quenching,
also commonly known as concentration quenching, is attributed to two main
processes: Er immiscibility in the host matrix and/or Er ion-ion interaction.
To delineate the origin of this photoluminescence reduction, we applied X-ray
absorption near-edge spectroscopy (XANES) and extended X-ray absorption fine
structure (EXAFS) analyses.
X-ray absorption near edge spectroscopy (XANES) study using the Er
LIII edge at 8358 eV confirmed that Er was in the optically active
trivalent state (Er3+), having an octahedral symmetry similar to Er
in Er2O3. No other chemical state was found up to at
least 14 at.%, indicating no formation of Er precipitate which is optically
inactive. To obtain further insight into the Er local coordination, three
different 4-Å local cluster models were constructed, corresponding to three
different possible Er configurations in the Y2O3 thin
film. In all three models, the first shell is O while the second shell can be
all Er3+ (first model), all Y3+ (second model), or a
mixture of Y3+ and Er3+ (third model). Because of the
almost perfect crystal structure match between Y2O3 and
Er2O3, the simulation of the Er-doped
Y2O3 local structure was simply accomplished by replacing
Y3+ with Er3+ in the Y2O3 lattice.
Shown in Figure 1a (left) is a pictorial view of the Er-doped
Y2O3 structure at
low Er concentration. In this case, the center absorbing Er has a second
coordination shell with a mixture of both Er and Y. At high Er concentration
where the alternating growth of Y2O3 and
Er2O3 resulting in an exsolution with
Er2O3-rich domains (Figure 1a, right), the Er local
environment is described by combining the first and third model.
Shown in Figure 1b are the k3-weighted EXAFS of the
Y2O3 thin films doped with 6 at.% Er, representing
samples with high photoluminescence yield. The best fit to the EXAFS using the
1st and 2nd model agreed fairly well with the EXAFS up to k ~ 6
Å-1, mainly from the first O shell, but failed to describe the
oscillations at higher k (Figure 1b top). This indicates that the
second nearest neighbors are neither all Er3+ in which case the
local environment of Er3+ would be similar to that of
Er3+ in Er2O3, nor all Y3+ which
would otherwise indicate an infinite dilution of Er3+ in
Y2O3. A best fit to the EXAFS was achieved with the
third model (Figure 1b bottom) when the Y:Er cation ratio is specified to be
3:1, as determined by XPS compositional analysis In this case, a coordination
number of 6 for the first O shell and 8-9 for the second shell were obtained.
Furthermore, Er3+ was found to be completely miscible in the
Y2O3 matrix up to at least 8 at.%, showing no evidence of
Y2O3 and Er2O3 phase segregation.
There is also no indication of Er-Er coordination within 4-Å proximity.
For the 14 at.% Er-doped Y2O3 thin film, representing
samples with low photoluminescence yield, a combination of the first (~60%) and
third model (~40%) best fitted the EXAFS spectrum (Figure 1c), with the Y:Er
cation ratio specified at 1:3, as determined by XPS. This is consistent with
the alternating RE-ALD process, resulting in a layer-like structure under these
deposition conditions. Since there is no indication of Er-Er coordination in
all samples doped with 6 to 14 at.% Er3+, it is concluded that the
photoluminescence quenching observed in samples with Er concentration exceeding
8 at.% is not due to Er immiscibility in Y2O3 but likely
due to Er ion-ion interaction. When the Er3+ concentration is
sufficiently small, the ions are evenly distributed in the
Y2O3 matrix with relatively large inter-ionic distances,
impeding ion-ion interaction. Consequently, the photoluminescence yield is
relatively high in the absence of these competing processes. As the
Er3+ concentration increases, there is more Er3+ within a
4-Å proximity of each other that they can interact, resulting in cooperative
energy upconversion or energy migration, leading to reduced photoluminescence
yield.
Using EXAFS, the origin of the observed concentration quenching of
photoluminescence for the Er3+:Y2O3 system was
delineated. The study also suggests that, in order to prevent ion-ion
interaction, no Er3+ should have another Er3+ as a second
nearest neighbor. This criteria sets an upper limit on the Er3+
concentration in the Y2O3 host at
~6x1021/cm3, or ~10 at.%, estimated by systematically
replacing Y3+ in the Y2O3 unit cell by
Er3+ while ensuring there is no direct Er-O-Er bonding. These
results are essential to the understanding of the Er3+ optical
properties in correlation to its local structure, allowing for the optimization
of the photoluminescence yield by controlling its distribution in the host
lattice.
Primary Citation
References
|