2D compound formation during copper dissolution: An electrochemical STM study

Article abstract of DOI:10.1016/j.susc.2006.01.109

The reversible formation of a 2D-CuI film on Cu(1 0 0) is studied by means of cyclic voltammetry in combination with electrochemical scanning tunneling microscopy. Exposing the Cu(1 0 0) electrode surface to an acidic and iodide containing electrolyte (5 mM H₂SO₄/1 mM KI) leads to the formation of a well ordered c(p × 2)-I adsorbate layer at potentials close to the onset of the anodic copper dissolution reaction. Copper dissolution starts at slightly more positive potentials preferentially at step edges in the presence of the iodide adlayer via the removal of copper material from kink sites at step edges. This increase of mobile Cu⁺ ions causes the local exceeding of the CuI solubility product (pKL = 11.3), thereby giving rise to the nucleation and growth of a laterally well ordered 2D-CuI film. Key structural motifs of the growing CuI film are closely related to the (1 1 1) plane of bulk CuI. Quite intriguing, the 2D-CuI film does not act as a passive layer. Copper dissolution proceeds even in the presence of this binary compound via an inverse step flow mechanism.

Full text of DOI:10.1016/j.susc.2006.01.109

Surface Science 600 (2006) 3971–3977
www.elsevier.com/locate/susc
2
D compound formation during copper dissolution:
An electrochemical STM study
*
P. Broekmann , N.T.M. Hai, K. Wandelt
Institut f u¨ r Physikalische und Theoretische Chemie, Universit a¨ t Bonn, Wegelerstrasse 12, D-53115 Bonn, Germany
Available online 6 May 2006
Abstract
The reversible formation of a 2D-CuI film on Cu(100) is studied by means of cyclic voltammetry in combination with electrochemical
scanning tunneling microscopy.
2 4
Exposing the Cu(100) electrode surface to an acidic and iodide containing electrolyte (5 mM H SO /1 mM KI) leads to the formation
of a well ordered c(p · 2)-I adsorbate layer at potentials close to the onset of the anodic copper dissolution reaction. Copper dissolution
starts at slightly more positive potentials preferentially at step edges in the presence of the iodide adlayer via the removal of copper mate-
+
rial from kink sites at step edges. This increase of mobile Cu ions causes the local exceeding of the CuI solubility product (pKL = 11.3),
thereby giving rise to the nucleation and growth of a laterally well ordered 2D-CuI film. Key structural motifs of the growing CuI film are
closely related to the (111) plane of bulk CuI. Quite intriguing, the 2D-CuI film does not act as a passive layer. Copper dissolution pro-
ceeds even in the presence of this binary compound via an inverse step flow mechanism.
ꢀ
2006 Elsevier B.V. All rights reserved.
Keywords: STM; Electrified interfaces; Corrosion; Anion effects
1
. Introduction
chemical deposition processes the mastering of copper-elec-
trolyte interfaces with or without additional potential con-
trol can be regarded as a particular challenge.
Copper as one of the key materials of the twenty-first
century has attracted tremendous attention over the last
two decades due to its nowadays realized application as
wiring material in the state-of-the-art chip design [1], here
replacing the former aluminium technology. However,
the on-going trend of miniaturization towards the nanome-
ter scale requires a more sophisticated understanding of the
relevant interface properties and processes of those devices
containing reactive materials like as copper. An atomic
scale understanding of corrosion phenomena, corrosion
inhibition by organics, oxidation, anodic dissolution and
the formation of passivating films on copper is thus of vital
interest and has consequently been in the focus of numer-
ous fundamental studies. Since modern processing lines
of chip fabrication also involves more and more ‘‘wet’’
The anodic Copper Dissolution Reaction (CDR) in
acidic electrolytes has been intensively studied in particu-
lar by Bard et al. [2,3] and Magnussen/Behm et al. [4–8]
using in situ STM. Active sites for the dissolution or
deposition are exclusively kinks at step edges as stated
by Stranski in 1920 [9]. Magnussen and co-workers elab-
orated that the local structure of kink sites can be signif-
icantly altered by the presence of specifically adsorbed
anions thus leading to changes in the overall surface reac-
tivity [4]. State-of-the-art video STM techniques even
allow the direct correlation of ‘‘global’’ parameters such
as the exchange current density on the one hand and
the ‘‘local’’ and microscopic dissolution rates at kink sites
on the other hand [6,7].
Substantial contributions to the microscopic under-
standing of the anodic CDR in alkaline electrolytes came
from Marcus/Strehblow et al. [10–13]. Quite contrary to
the behavior in acidic media the CDR under alkaline
*
Corresponding author. Tel.: +49 228 73 3292; fax: +49 228 73 2551.
E-mail address: broekman@thch.uni-bonn.de (P. Broekmann).
0
039-6028/$ - see front matter ꢀ 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2006.01.109

3972
P. Broekmann et al. / Surface Science 600 (2006) 3971–3977
conditions is interfered by a binary compound formation,
namely the formation of a passivating Cu O film at lower
2
potentials and a Cu O/CuO, Cu(OH) duplex film at high-
2
2
er potentials. Experimental conditions such as the pH, the
applied potential and the polarization time sensitively con-
trol the atomic scale ordering, the overall film morphology
and the thickness of the formed passive layers [13].
In the present study we report a new dissolution/corro-
sion mechanism observed on a Cu(100) electrode surface
which comprises characteristic elements of the dissolution
mechanism observed in both acidic and alkaline electro-
lytes. A binary compound formation is observed even
in acidic electrolytes when the solution phase contains
strongly adsorbing species like iodide which is known to
Fig. 1. Cyclic voltammograms of Cu(100) in pure 5 mM H
2 4
SO (black
curve) and in 5 mM H SO /1 mM KI (grey curve), dE/dt = 10 mV/s.
2
4
+
form insoluble salts with Cu ions. The main focus of the
present paper lies in the structural characterization of a
well ordered 2D-CuI layer which forms and, by contrast
deviations are related to the appearance of a pronounced
anodic peak system with P1 at +175 mV and P2/3 centered
at about +215 mV. In the reverse potential scan character-
to the Cu O/CuO film, does not passivate the electrode
2
0
against on-going CDR.
istic cathodic current waves emerge at +120 mV (P 1),
0
0
ꢀ
110 mV (P 2) and +40 mV (P 3). The lack of an exponen-
tially increasing anodic current, at least within the given
potential range, is clearly indicative for a passivation of
the copper surface with respect to the CDR after passing
the anodic peak system P1–P3. An exponential increase
of the CDR is only observed when potentials above
2
. Experimental
In order to study the initial stage of anodic CDR in the
presence of inorganic salts we use a home-built in situ STM
14].
The tunneling tips were electrochemically etched in 2M
[
+
360 mV are applied [16]. Note that the pure supporting
electrolyte (black curve in Fig. 1) the CDR starts already
at about +275 mV.
KOH solution from 0.25 mm tungsten wire and subse-
quently coated with commercially available hot glue.
For all solutions high purity water (Milli-Q purification
system; conductivity < 18 MX cm; TOC < 5 ppb) and re-
agent grade chemicals were used. Given the high affinity
of copper towards oxygen the electrolyte had to be de-
gassed with argon for several hours before use. All poten-
tials given in the text refer to a RHE reference electrode.
Prior to each experiment to the copper surface had to be
treated by an electrochemical etching procedure as de-
scribed in Ref. [14].
A similar electrochemical behavior as presented in
Fig. 1 has been reported by Inukai et al. [17] for a
Cu(111) surface exposed to an iodide containing per-
chloric acid. The additional anodic current features were
explained in terms of copper oxidation in the presence
of iodide anions leaving CuI behind. In principle the same
explanation as given by Inukai et al. [17] holds for the
Cu(100) electrode in the iodide containing sulfuric acid.
Recent SXPS (Synchrotron X-ray Photoelectron Spectros-
copy) studies, indeed, evidenced the formation of CuI
after passing the anodic peak system. The discussion of
the particular reaction mechanism will be addressed in a
separate paper [25].
All experiments started routinely with the characteriza-
tion of the Cu(100) electrode in the presence of pure
5
mM H SO which served as supporting electrolyte for
2 4
all subsequent dissolution experiments in the presence of
iodide anions. For the latter experiments the pure support-
ing electrolyte was exchanged under potential control, e.g.
at E = +80 mV, by a mixture of 5 mM H SO and 1 mM
It is, actually, this thick insulating film of CuI which
leads to the observed passivation of the copper electrode
with respect to the CDR. The large separation/hysteresis
of about 320 mV between the oxidation (P2) and the corre-
2
4
KI solution. The ‘‘intermediate step’’ via the pure sulfuric
acid is important in order to avoid the instantaneous for-
mation thick CuI films.
0
sponding reduction peak (P 2) points to a strong kinetic
hindrance in particular of the reduction processes [16].
With this pronounced passivation behavior the Cu/I sys-
tem seems to be closer to the electrochemical behavior of
Cu in alkaline electrolytes [13] than to other acidic electro-
lytes [4,15,16]. A detailed analysis of all subtle current fea-
tures showing up in Fig. 1 will be given in forthcoming
papers [16,25]. Here we restrict ourselves to the surface
dynamics occurring within the potential range between
+80 mV and +130 mV at the onset of the anodic peak sys-
tem where copper oxidation remains restricted to the
monolayer regime only.
3
. Results and discussion
3
.1. Electrochemistry
A representative cyclic voltammogram of Cu(100) in
mM H SO /1 mM KI solution is presented in Fig. 1 (grey
5
2
4
curve). It reveals a number of striking deviations to other
CVs of Cu(100) in various acidic electrolytes [4,15]. These

P. Broekmann et al. / Surface Science 600 (2006) 3971–3977
3973
3
.2. STM results
of local dissolution processes (Fig. 3(a) and (b)) in the pres-
ence of the c(p · 2)-I overlayer. Local dissolution starts at
step edges and proceeds via an ‘‘inverse step-flow’’ mecha-
nism. Steps stabilize themselves during dissolution along
preferential directions (Fig. 3(b)) which coincide with main
symmetry axes of the iodide lattice [18]. Besides the local
removal of copper material one observes the 2D growth
of an ordered film in close proximity of the ongoing copper
dissolution (Fig. 3(b)). For two reasons it can be excluded
that the observed film growth is a simple re-deposition of
afore dissolved copper material: firstly, the emergence of
a striped long-range superstructure does not correspond
to the c(p · 2)-I overlayer. Secondly, the growth of a com-
pletely new phase is inferred from one-dimensional cross
sections over the growing film in the sub-monolayer regime
(Fig. 3(c) and (d)). Step heights do not correspond to the
value of 0.18 nm expected for step heights of the bare or
anion covered Cu(100) surface. Using the height difference
between two iodide covered Cu(100) terraces as internal
reference (Fig. 3(c)) one determines a height difference of
0.39 ± 0.01 nm between an iodide covered (100) terrace
and the growing film (Fig. 3(d)) which is actually even
more than twice the mono-atomic step height on
Cu(100). The height difference between two adjacent ter-
races both covered by this new phase amounts to 0.18 nm
An electrolyte exchange procedure at +80 mV as de-
scribed in the experimental section leads instantaneously
to the formation of a laterally well ordered iodide adlayer
(
Fig. 2(a)). Iodide anions form a distorted ‘‘pseudo’’-
square structure on Cu(100) [18,19] which is commensu-
rate in only one of the substrate h011i directions giving rise
to an additional one-dimensional height modulation per-
pendicular to this ‘‘commensurate direction’’. To first
approximation we can understand this structure in terms
of a distorted c(2 · 2) lattice which is expanded parallel
to only one of the original c(2 · 2) unit-cell vectors leading
to a c(p · 2) unit cell. Characteristically, this p-vector de-
creases (increases) with increasing (decreasing) electrode
potentials in terms of a so-called electrocompression
(
electrodecompression) process [18,19]. A schematic hard-
sphere model of the c(p · 2) is presented in Fig. 2(c). At
the given potential of +80 mV (Fig. 2(a)) the iodide satura-
tion coverage of H = 0.4 ML is already reached corre-
sponding to p = 2.5 and an NND (nearest neighbor
distance) of 0.409 nm. Similar electrocompressible iodide
phases are reported for the Cu(111) electrode [24].
Sweeping the electrode potential into positive direction
from E = +80 mV to E = +100 mV causes the initiation
Fig. 2. (a) Structure of the c(p · 2)-I layer at the onset of the CDR, 15.1 nm · 15.1 nm, I
t
= 5 nA, Ubias = 1 mV, Ework = +80 mV; (b) 4.7 · 4.7 nm,
I
t
= 5 nA, Ubias = 1 mV, Ework = +80 mV; (c) Hard sphere model of the c(p · 2)-I layer, the p-value amounts to p = 2.5 at Ework = +80 mV.

3974
P. Broekmann et al. / Surface Science 600 (2006) 3971–3977
Fig. 3. Simultaneously occurring copper dissolution and 2D-CuI film growth, (a) 108 nm · 108 nm, I
08 nm · 108 nm, I = 0.25 nA, Ubias = 224 mV, Ework = +120 mV; (c) cross section along the marked white line in (b); (d) cross section along the marked
white line in (b).
t
= 0.25 nA, Ubias = 224 mV, Ework = +110 mV; (b)
1
t
which reflects again a single Cu–Cu step. Characteristically,
the 2D film growth starts within a narrow potential range
of only 10 mV between +100 mV and +110 mV at the on-
set of the anodic peak system P1/P2 and proceeds until the
entire surface is covered with this new phase. One reason-
able explanation for all these observations is the formation
of an ultra-thin 2D film of CuI. Dissolved copper material
does not leave the iodide covered copper surface into the
solution but increases the local concentration of copper
ions directly at the surface, with the consequence that the
low solubility product of CuI (pKL = 11.3) becomes
locally exceeded. This actually gives rise to the heteroge-
neous nucleation and growth of a surface confined 2D
compound, namely CuI, as argued in the following.
Since the usual STM imaging mode does not provide
any chemical sensitivity our assignment to 2D-CuI is based
on a comparison to the work by Andryushechkin et al. [21]
dealing with characterization of CuI, obtained after expos-
ing Cu(100) to an excess of molecular iodine under UHV
conditions. In fact, these authors observed with STM the
same sequence of iodine phases and CuI films as a function
of iodine exposure in UHV [20–23] as we have seen by
in situ STM with a Cu(100) electrode under electrochemi-
cal conditions. Beyond this ‘‘gross’’ similarity there is even
a striking correspondence between the atomic structure of
our supposed CuI film (Fig. 4) and that grown under
UHW conditions. This latter UHV-grown film could, in-
deed, be identified as CuI using Auger electron spectros-
copy [21].
The 2D-CuI phase presented in Fig. 4 is characterized
by stripes arranged in a strictly parallel manner with two
distinct
periodicities
of
d1 = 1.7 ± 0.2 nm
and
d2 = 3 ± 0.2 nm. However, there is no apparent regularity
in the sequence of these characteristic stripe distances. The
almost ideal hexagonal symmetry of the 2D-CuI phase be-
comes obvious from Fig. 4(b). The averaged NND amounts
to 0.44 ± 0.02 nm. Local deviations from that value are
due to the existence of periodically arranged line disloca-
tions. Close packed rows in adjacent stripes exhibit a pro-
nounced anti-phase behavior as indicated by the dashed
lines in Fig. 4(b). The enclosing angle between the propaga-
tion direction of the stripe pattern and the close packed
rows on the atomic scale amounts to 8 ± 2ꢁ [16]. All these
observations are in excellent agreement with the assump-

P. Broekmann et al. / Surface Science 600 (2006) 3971–3977
3975
Fig. 4. Atomic structure of the stable 2D-CuI film, (a) 27.3 nm · 27.3 nm, I
bias = 10 mV, Ework = +120 mV.
t
= 1.3 nA, Ubias = 10 mV, Ework = +120 mV; (b) 3 nm · 3 nm, I
t
= 1.3 nA,
U
tion of a CuI(111) phase formed on the iodide pre-covered
Cu(100) surface. Bulk CuI crystallizes in a ZnS-type struc-
ture (zinc blende) with a tetrahedral coordination (KZ = 4)
of Cu and I respectively. Note, the formal oxidation
states (+I) for copper and (ꢀI) for iodide, however, do
not imply an ionic bonding character. Quite in contrast,
all solid monovalent copper halide phases reveal a highly
covalent bonding between copper and halide particles
which is actually the chemical reason for the low solubility
of these compounds.
distorted tetrahedron with a central monovalent Cu species
surrounded by four iodide particles as a characteristic
structure motif of the growing 2D-CuI film. However,
due to the different lattice symmetries and the slight mis-
match between nearest neighbor spacings in the c(p · 2)-I
phase (NNDiodide = 0.41 nm) and an ideal CuI(111) plane
(NNDiodide = NNDCu = 0.422 nm), respectively, those lat-
tices are apparently not in full registry. The appearance
of striped dislocation lines (Fig. 4) in combination with
the pronounced anti-phase behavior could be explained
in terms of a partial release of strain which originates from
the inherent lattice mismatch. It should be stressed, how-
ever, that an ultimate structure model including these ‘‘bur-
ied’’ interfaces cannot be achieved only on the basis of
STM work. Forthcoming in situ X-ray diffraction experi-
ments on this system will certainly help to improve our
preliminary structure model.
+
ꢀ
A preliminary model for the 2D-CuI surface film is
based on the following stacking sequence: Cu(100)/
c(p · 2)-I/Cu/I (Fig. 5) which is in agreement with the re-
sults of Andryushechkin et al. [21]. Due to the reduced
symmetry of the c(p · 2)-I adsorbate phase with respect
to the underlying substrate (Fig. 2) there are also somewhat
distorted ‘‘quasi’’-threefold hollow sites available within
the iodide adsorbate layer as potential adsorptions sites
One of the most prominent properties of this 2D-CuI
film is related to the on-going CDR at slightly more posi-
tive potentials (+120 mV). It appears quite intriguing that
the 2D-CuI film does not effectively passivate the electrode
+
for Cu ions (Fig. 2(c)). With an additional iodide particle
+
on-top of this single Cu ion it is possible to form a slightly
surface against the CDR as it is reported for the Cu O/
2
CuO duplex film in alkaline solution [12,13]. Fig. 6 demon-
strates an on-going dissolution reaction in the presence of
the 2D-CuI film. Time resolved STM images clearly indi-
cate an inverse step flow mechanism with receding step
edges leaving the same CuI film at the lower terraces be-
hind. Considering the stacking sequence at the interface
(Fig. 5) it becomes obvious that the dissolution mechanism
in the presence of the 2D-CuI film must be much more
complex than for electrode surfaces which are covered by
monolayers of regular adsorbate phases [4,8]. In this latter
case only two layers are involved into the dissolution reac-
tion, namely the anion overlayer and the metallic substrate
layer underneath while in the present case four layers must
participate in the ‘‘collective receding’’ of steps (Fig. 6).
This concerted process is indicated in Fig. 5 by four dark
arrows.
Fig. 5. Schematic drawing illustrating the vertical stacking sequence of
Cu(100)/I/Cu/I system; this model explains the observed step height in
Fig. 3(b).

3976
P. Broekmann et al. / Surface Science 600 (2006) 3971–3977
t
Fig. 6. CDR in the presence of the 2D-CuI film, (a)–(c) 70 nm · 70 nm, I = 1 nA, Ubias = 5 mV, Ework = +120 mV, one prominent defect within the 2D-
CuI film is marked in all 4 STM images, the in-set in panel (c) shows the structure of the 2D-CuI film which can be imaged on terraces affected by the
CDR.
Increasing the dissolution rate by applying potentials
above +130 mV leads to the formation of 3D-CuI/iodine
clusters [16,25] until the entire surface is covered with an
about 5–8 nm thick passive film.
the solution phase while in our case the metal cations orig-
inate directly from the electrode surface as a consequence
of an oxidative dissolution reaction of the electrode mate-
rial. The observed structural features are in good agree-
ment with the assumption of a Cu(100)/I/Cu/I stacking
sequence. Due to the pronounce hexagonal symmetry of
the growing film we propose a lateral structure of the
CuI film which is closely related to the (111)-plane of bulk
4
. Conclusions
In this paper we could show that a well defined binary
compound is formed at the electrode surface close to
the onset of the copper dissolution reaction (CDR). An
+
CuI with a ‘‘pseudo’’-tetrahedral coordination of Cu ions
placed in a kind of ‘‘sandwich’’ of two adjacent iodide lay-
ers. Quite surprisingly, the formation of the 2D-CuI film
does not efficiently suppress the anodic dissolution of cop-
per. Only the ongoing CDR in the presence of the 2D-CuI
film at higher reaction rates leads to the growth of 3D-CuI
clusters thus explaining the final passivation against the
CDR as observed in the CV.
ultimate prerequisite for such
a binary compound
formation is the extraordinarily low solubility of this
binary compound.
A laterally well ordered 2D-CuI film is formed on
Cu(100) at potentials (+100 mV to +110 mV) below the
potential range of the corresponding bulk CuI formation
(
above +130 mV). Due to this separation of the potential
regime of the 2D-CuI formation on the one hand and the
D-CuI bulk formation on the other we denote the ob-
References
3
served films growth as Under Potential Formation (UPF)
of a binary compound similar to the well known phenom-
enon of Under Potential Deposition (UPD) of metals. In
this latter case, however, the metal ‘‘deposit’’ comes from
[
1] T.G. Koetter, H. Wendrock, H. Schuehrer, C. Wenzel, K. Wentzig,
Microelectron. Reliab. 40 (2000) 1295.
[2] D.W. Suggs, A.J. Bard, J. Am. Chem. Soc. 116 (1994) 10725.
[3] D.W. Suggs, A.J. Bard, J. Phys. Chem. 99 (1995) 8349.

P. Broekmann et al. / Surface Science 600 (2006) 3971–3977
3977
[
4] M.R. Vogt, A. Lachenwitzer, O.M. Magnussen, R.J. Behm, Surf. Sci.
99 (1998) 49.
[16] P. Broekmann, N.T.M. Hai, K. Wandelt, J. Appl. Electrochem.,
in press.
3
[
[
5] O.M. Magnussen, M.R. Vogt, Phys. Rev. Lett. 85 (2000) 357.
6] O.M. Magnussen, L. Zitzler, B. Gleich, M.R. Vogt, R.J. Behm,
Electrochem. Acta 46 (2001) 3725.
[17] J. Inukai, Y. Osawa, K. Itaya, J. Phys. Chem. B 102 (1998) 10034.
[18] P. Broekmann, A. Spaenig, A. Hommes, K. Wandelt, Surf. Sci. 517
(1–3) (2002) 123.
[
7] O.M. Magnussen, W. Polewska, L. Zitzler, B. Gleich, R.J. Behm,
Faraday Discuss. 121 (2002) 43.
[19] A. Hommes, A. Spaenig, P. Broekmann, K. Wandelt, Surf. Sci. 547
(2003) 239.
[
[
8] O.M. Magnussen, Chem. Rev. 102 (2002) 679.
9] I.N. Stranski, Z. Phys. Chem. 136 (1928) 259.
[20] B.V. Andryushechkin, K.N. Eltsov, V.M. Shevlyuga, U. Bardi, B.
Cortigiani, Surf. Sci. 497 (2002) 59.
[
[
10] V. Maurice, H.-H. Strehblow, P. Marcus, Surf. Sci. 458 (2000) 185.
11] V. Maurice, L.H. Klein, H.-H. Strehblow, P. Marcus, J. Electrochem.
Soc. 150 (2003) B316.
12] J. Kunze, V. Maurice, L.H. Klein, H.-H. Strehblow, P. Marcus, J.
Electroanal. Chem. 554–555 (2003) 113.
13] J. Kunze, V. Maurice, L.H. Klein, H.-H. Strehblow, P. Marcus, Corr.
Sci. 46 (2004) 245.
14] M. Wilms, M. Kruft, G. Bermes, K. Wandelt, Rev. Sci. Instr. 70 (7)
[21] B.V. Andryushechkin, K.N. Eltsov, V.M. Shevlyuga, Surf. Sci.
566–568 (2004) 203.
[22] B.V. Andryushechkin, K.N. Eltsov, V.M. Shevlyuga, Surf. Sci. 472
(2001) 80.
[23] B.V. Andryushechkin, K.N. Eltsov, V.M. Shevlyuga, Surf. Sci. 584
(2005) 287.
[24] B. Obliers, P. Broekmann, K. Wandelt, J. Electroanal. Chem.
554–555 (2003) 183.
[
[
[
[
(1999) 3641.
[25] P. Broekmann, N.T.M. Hai, S. H u¨ mann, R. Hunger, K. Wandelt,
in preparation.
15] P. Broekmann, M. Anastasescu, A. Spaenig, W. Lisowski, K.
Wandelt, J. Electroanal. Chem. 500 (2001) 241.