86
4 Ultrathin Layers
on Se-covered electrode was studied by oxidative stripping of Cd (reaction: Cd
Cd
2+
+ 2e). The quantitative analysis of the Se UPD on Cd, however, required the
reductive stripping of Se (reaction: Se + 2e Se
2− ) because the oxidative stripping
would have led to the oxidation of the Cd layer, too. The width of the potential
interval where a monolayer could be produced varied from one component to the
other, yielding intervals of about 90 and 40 mV width for Cd and Se deposition,
respectively. It seems to be a general experience that the optimal potential interval
for the deposition of the metallic component is wider than that of the chalcogenide
component. For EC-ALD of bismuth telluride, Zhu et al. found [28] a nearly 300 mV
electrode potential range for the monolayer formation of Bi, while Te deposition with
an appropriate coverage required a more accurate choice of potential within a ~70 mV
potential interval. When CuS is deposited by EC-ALD, the Cu deposition potential
can be chosen in a 180 mV potential interval [29].
The choice of the deposition potentials can be also verified by observing the
stoichiometry of the deposit. If the desired ratio of the element is found, the deposition
potentials can be accepted. As shown for CdTe deposition [30], the stoichiometry of
the deposit, observed as a function of the Cd deposition potential, showed a plateau
region in the same Cd deposition potential range where the coverage also had a
plateau as a function of the same independent variable, although the coverage per
cycle for Cd in the optimal potential interval was less than one monolayer.
It is to be stressed that the optimization of the UPD deposition potentials of atomic
monolayers or bilayers on a noble metal substrate (i.e. normally used in preliminary
experiments) is not a guarantee that the deposition of successive layers (bilayers)
can take place in a similar way at large cycle numbers. The explanation is that the
properties of the first bilayer on a foreign substrate can be substantially different
from the surface of a bulk deposit. As described by Wade et al. [31] for InAs and
InSb, the application of the same deposition potentials that result in a well-defined
monolayer on a noble metal electrode may lead to the stop of the entire deposition
process later. In contrast, if the deposition potentials are chosen by 200–300 mV more
negative than the potentials of the UPD range on the starting substrate, the deposit
keeps growing in the subsequent cycles at a reasonable rate. In such a case, however,
excessive deposition currents may occur in the first few tens of cycles due to the initial
roughening that is related to the three-dimensional nucleation instead of a mere UPD
at the beginning. The deposition potentials chosen from the cyclic voltammograms
run on the bare substrate are often not sufficiently negative to achieve the growth of
the deposit with large cycle numbers. This points out that “the potentials derived from
voltammetry on the substrate are simply a good first approximation for the potentials
needed for at most one cycle, but do not represent the potentials needed to deposit
the elements onto each other” [32]. In many cases, especially for multicomponent
structures, it appears that there is no potential set that would work both in the initial
phase and in the steady-state [33].
It is not only the chemical nature of the surface that can lead to a change in
the deposition potential suitable to produce full monolayer coverage. Since most of
the deposits prepared by EC-ALD are semiconductors, the potential drop across the
semiconductor layer of increasing thickness also varies because of the occurrence
4 Ultrathin Layers
on Se-covered electrode was studied by oxidative stripping of Cd (reaction: Cd
Cd
2+
+ 2e). The quantitative analysis of the Se UPD on Cd, however, required the
reductive stripping of Se (reaction: Se + 2e Se
2− ) because the oxidative stripping
would have led to the oxidation of the Cd layer, too. The width of the potential
interval where a monolayer could be produced varied from one component to the
other, yielding intervals of about 90 and 40 mV width for Cd and Se deposition,
respectively. It seems to be a general experience that the optimal potential interval
for the deposition of the metallic component is wider than that of the chalcogenide
component. For EC-ALD of bismuth telluride, Zhu et al. found [28] a nearly 300 mV
electrode potential range for the monolayer formation of Bi, while Te deposition with
an appropriate coverage required a more accurate choice of potential within a ~70 mV
potential interval. When CuS is deposited by EC-ALD, the Cu deposition potential
can be chosen in a 180 mV potential interval [29].
The choice of the deposition potentials can be also verified by observing the
stoichiometry of the deposit. If the desired ratio of the element is found, the deposition
potentials can be accepted. As shown for CdTe deposition [30], the stoichiometry of
the deposit, observed as a function of the Cd deposition potential, showed a plateau
region in the same Cd deposition potential range where the coverage also had a
plateau as a function of the same independent variable, although the coverage per
cycle for Cd in the optimal potential interval was less than one monolayer.
It is to be stressed that the optimization of the UPD deposition potentials of atomic
monolayers or bilayers on a noble metal substrate (i.e. normally used in preliminary
experiments) is not a guarantee that the deposition of successive layers (bilayers)
can take place in a similar way at large cycle numbers. The explanation is that the
properties of the first bilayer on a foreign substrate can be substantially different
from the surface of a bulk deposit. As described by Wade et al. [31] for InAs and
InSb, the application of the same deposition potentials that result in a well-defined
monolayer on a noble metal electrode may lead to the stop of the entire deposition
process later. In contrast, if the deposition potentials are chosen by 200–300 mV more
negative than the potentials of the UPD range on the starting substrate, the deposit
keeps growing in the subsequent cycles at a reasonable rate. In such a case, however,
excessive deposition currents may occur in the first few tens of cycles due to the initial
roughening that is related to the three-dimensional nucleation instead of a mere UPD
at the beginning. The deposition potentials chosen from the cyclic voltammograms
run on the bare substrate are often not sufficiently negative to achieve the growth of
the deposit with large cycle numbers. This points out that “the potentials derived from
voltammetry on the substrate are simply a good first approximation for the potentials
needed for at most one cycle, but do not represent the potentials needed to deposit
the elements onto each other” [32]. In many cases, especially for multicomponent
structures, it appears that there is no potential set that would work both in the initial
phase and in the steady-state [33].
It is not only the chemical nature of the surface that can lead to a change in
the deposition potential suitable to produce full monolayer coverage. Since most of
the deposits prepared by EC-ALD are semiconductors, the potential drop across the
semiconductor layer of increasing thickness also varies because of the occurrence