92
4 Ultrathin Layers
SEM images recorded for EC-ALD layers indicate that although the deposition
method itself was designed as a layer-by-layer process, the nature of the deposit
turns into a granular one as the cycle number increases to a few dozens or more.
This trend was observed for a variety of layer composition, even if the substrate is
a well-defined single crystal face [8, 9, 51]. The deviation from the planar layer-bylayer nature of the process can be understood by considering the stress originating
from the misfit between the atomic structure of the substrate surface and that of
the relaxed structure of the deposit. The granular growth sounds even more natural
if the substrate is polycrystalline [11, 15, 22, 28, 36, 37]. The morphology of the
deposit is also sensitive to the composition of the solutions applied for the deposition
processes [79]. When the substrate does not allow a uniform nucleation of the ECALD coating, the deposit may be intentionally granular [80] or dendritic [55]. In
both latter cases, the process seems to begin in a similar manner to an instantaneous
nucleation since the size distribution of the resulting features is narrow. This results
in a situation that even if the growth starts on a porous material like a nanowire array,
the size of the deposit will be similar, regardless of the longitudinal position of the
EC-ALD-produced dot on the nanowire [81]. This is well in accord with the nature
of the atom-by-atom deposition process.
The granular nature of the deposit can obtain a quantitative assessment with a
height distribution of the deposit if AFM is used for surface imaging [21, 29, 44].
AFM is also suitable to characterize the evolution of the surface roughness with cycle
number. It proved to be common that the surface roughness starts increasing from the
early stage of the growth process. Depending on the deposit, the surface roughness
may either tend to saturate after a few hundred cycles [59] or keep increasing without
saturation [82]. The way as the surface roughness increases upon growth also depends
on the deposition potential of each atomic layer [42]. AFM was also used to keep
track of the change in the surface morphology of EC-ALD deposits after exposure
to air due to oxidation [73].
The major goal of depositing EC-ALD layers is to exploit their semiconductor
properties. As a primary parameter, the band gap of the semiconductor layer deposited
has been determined for a number of samples [23, 30, 32, 34, 37–39, 41, 48, 71, 72,
80]. The determination of the band gap is based on the following equation:
α(hν)hν = A(hν − E G )
n
(4.1)
where h is the Planck constant, ν is the photon frequency, α is the absorption coefficient at the photon energy specified in the argument, E G is the energy gap and A is
a proportionality constant. The exponent n is equal to ½ for a direct band gap and
2 for an indirect band gap. The graphical method of the determination of the direct
band gap is shown in Fig. 4.6.
The experience related to the band gap found for thick EC-ALD semiconductor
layers is that it approaches the bulk value (or that measured for single-crystalline
reference samples). However, at small layer thicknesses, a blue shift is often found
[39], which is due to the quantum confinement.
4 Ultrathin Layers
SEM images recorded for EC-ALD layers indicate that although the deposition
method itself was designed as a layer-by-layer process, the nature of the deposit
turns into a granular one as the cycle number increases to a few dozens or more.
This trend was observed for a variety of layer composition, even if the substrate is
a well-defined single crystal face [8, 9, 51]. The deviation from the planar layer-bylayer nature of the process can be understood by considering the stress originating
from the misfit between the atomic structure of the substrate surface and that of
the relaxed structure of the deposit. The granular growth sounds even more natural
if the substrate is polycrystalline [11, 15, 22, 28, 36, 37]. The morphology of the
deposit is also sensitive to the composition of the solutions applied for the deposition
processes [79]. When the substrate does not allow a uniform nucleation of the ECALD coating, the deposit may be intentionally granular [80] or dendritic [55]. In
both latter cases, the process seems to begin in a similar manner to an instantaneous
nucleation since the size distribution of the resulting features is narrow. This results
in a situation that even if the growth starts on a porous material like a nanowire array,
the size of the deposit will be similar, regardless of the longitudinal position of the
EC-ALD-produced dot on the nanowire [81]. This is well in accord with the nature
of the atom-by-atom deposition process.
The granular nature of the deposit can obtain a quantitative assessment with a
height distribution of the deposit if AFM is used for surface imaging [21, 29, 44].
AFM is also suitable to characterize the evolution of the surface roughness with cycle
number. It proved to be common that the surface roughness starts increasing from the
early stage of the growth process. Depending on the deposit, the surface roughness
may either tend to saturate after a few hundred cycles [59] or keep increasing without
saturation [82]. The way as the surface roughness increases upon growth also depends
on the deposition potential of each atomic layer [42]. AFM was also used to keep
track of the change in the surface morphology of EC-ALD deposits after exposure
to air due to oxidation [73].
The major goal of depositing EC-ALD layers is to exploit their semiconductor
properties. As a primary parameter, the band gap of the semiconductor layer deposited
has been determined for a number of samples [23, 30, 32, 34, 37–39, 41, 48, 71, 72,
80]. The determination of the band gap is based on the following equation:
α(hν)hν = A(hν − E G )
n
(4.1)
where h is the Planck constant, ν is the photon frequency, α is the absorption coefficient at the photon energy specified in the argument, E G is the energy gap and A is
a proportionality constant. The exponent n is equal to ½ for a direct band gap and
2 for an indirect band gap. The graphical method of the determination of the direct
band gap is shown in Fig. 4.6.
The experience related to the band gap found for thick EC-ALD semiconductor
layers is that it approaches the bulk value (or that measured for single-crystalline
reference samples). However, at small layer thicknesses, a blue shift is often found
[39], which is due to the quantum confinement.