PHYSICAL REVIEW B 66, 115401 共2002兲

Observation of preferred heights in Pb nanoislands: A quantum size effect

Roberto Otero,* Amadeo L. Vázquez de Parga, and Rodolfo Miranda
Departemento de Fı́sica de la Materia Condensada and ‘‘Instituto Nicolás Cabrera,’’ Universidad Autónoma de Madrid,
28049 Madrid, Spain
共Received 5 July 2002; published 3 September 2002兲
Scanning-tunneling microscopy observation of Pb nanocrystals grown on Cu共111兲 indicates that in the
equilibrium distribution of island heights some heights appear much more frequently than others. The simul-
taneous, local spectroscopic characterization allows us to relate this finding to the confinement of electrons
within the islands: the ‘‘magic,’’ preferred heights correspond to islands having a quantum well state far from
the Fermi energy, while the ‘‘forbidden’’ heights turn out to be those that would have a quantum well state
close to the Fermi level. The ‘‘magic’’ and ‘‘forbidden’’ heights can be predicted from the Fermi-surface
geometry of bulk Pb.

DOI: 10.1103/PhysRevB.66.115401 PACS number共s兲: 73.21.Fg, 61.46.⫹w, 68.55.Jk

I. INTRODUCTION ated to this particular film thickness has a significant contri-

bution to the stability of the Ag film.
Spatial confinement of electrons by a one-dimensional In this paper we report on the decisive influence of
共1D兲 potential well results in discrete energy states, known as QWS’s to dictate the heights observed in the equilibrium
quantum well states 共QWS’s兲. By adjusting the physical di- distribution of nanometer-high Pb islands grown on Cu共111兲.
mensions of the confinement region, QWS’s can be placed at We explore a wide range of Pb coverage from 4 equivalent
the Fermi level 关with the corresponding increase in the den- ML up to 22 equivalent ML. Both the height of the islands
sity of states 共DOS兲兴. This is important because the DOS at and their electronic structure are locally determined with
the Fermi level triggers a number of electronic phase transi- scanning-tunneling microscopy 共STM兲. We will demonstrate
tions 共superconductivity, charge-density waves, lattice distor- that the island heights most probable are those presenting a
tions, ferromagnetism, antiferromagnetism兲. For example, QWS placed far from the Fermi level, while those that only
the 共spin-polarized兲 QWS’s play a decisive role in the oscil- rarely appear have a QWS located close to the Fermi level.
latory coupling of magnetic layers across nonmagnetic, me- In other words, the islands rearrange their height to avoid the
tallic spacers.1,2 In magnetic/nonmagnetic trilayers, it has presence of QWS’s at the Fermi level. Furthermore, we will
been found that when the nonmagnetic spacer thickness is show that the shape of the Fermi surface of Pb and, in par-
such that there would be a maximum in the DOS at the ticular, the nesting vectors, determine the distribution of
Fermi level due to a QWS, the second magnetic layer orders heights found in our experiments.
antiferromagnetically with respect to the first one.2 This low-
ers the symmetry of the system and eliminates the QWS at II. EXPERIMENTAL PROCEDURES
the Fermi energy, thereby decreasing the total energy of the
system. Thus, one can say that the antiferromagnetic cou- The experiments have been carried out in an ultrahigh
pling is driven by the need to avoid the presence of a QWS vacuum chamber containing a homemade STM, rear-view
at the Fermi level. low-energy electron-diffraction optics, Auger spectroscopy,
Due to the electronic states produced by the confinement, and facilities for ion sputtering, cooling the sample and de-
the total electronic energy will oscillate with the thickness of positing metals. The Cu共111兲 single crystal was electro-
metallic films. If the contribution of the electronic energy to chemically polished, sputter-annealed 共500 eV, 800 K兲 for 5
the total energy is significant, some sizes could be energeti- min, and cooled down to 300 K. Pb was evaporated by elec-
cally more favorable than others. In fact, the energy associ- tron bombardment with a deposition rate of 0.1 ML/min with
ated to the confined motion of electrons was postulated to be the sample at room temperature. After the deposition the
important for the appearance of ‘‘magic’’ sizes in the forma- sample was flash-annealed to 400 K. The STM tips were
tion of Na clusters3 in free space. The possibility of quantum made of electrochemically etched polycrystalline W and
size effects affecting morphology in thin films has also been cleaned and characterized for spectroscopic measurements as
reported for metallic islands on semiconductor substrates, described elsewhere.11
i.e., a single ‘‘magic’’ height for Ag films on GaAs共110兲,4
and Ag 共Refs. 5 and 6兲 and Pb 共Refs. 7 and 8兲 on Si共111兲. It III. STATISTICAL ANALYSIS OF ISLAND HEIGHTS
has been proposed that the competition between quantum
confinement, charge spilling, and interface-induced Friedel Depositing Pb on Cu共111兲 at 300 K and above12 leads to a
oscillation defined the existence of critical/magical thick- Stranski-Krastanov mode of growth,13 i.e., after the comple-
nesses in these cases.9 Recently, it has been reported that tion of the first monolayer 3D Pb islands grow on the sur-
5-monolayer 共ML兲-thick Ag films on Fe共100兲 are especially face. Figure 1 shows a representative image of this mode of
stable.10 In the theoretical model used it was concluded that growth. The characteristic lateral dimension of the islands is
the contribution of the electronic energy of the QWS associ- around 400– 600 Å for our growth conditions. The islands

0163-1829/2002/66共11兲/115401共6兲/$20.00 66 115401-1 ©2002 The American Physical Society

OTERO, VÁZQUEZ DE PARGA, AND MIRANDA PHYSICAL REVIEW B 66, 115401 共2002兲

FIG. 1. STM topography of an area 3000⫻3000 Å2 showing

several Pb islands on a stepped Cu共111兲 surface. The image was
taken at a sample bias of 1.2 V. The top surface of the Pb crystallites
is atomically flat, but the steps of the substrate can still be recog-
nized at the surface of the islands due to the difference in lattice
parameter between Pb and Cu. The lateral dimension of the islands
is 400– 600 Å.

are 共111兲 oriented Pb nanocrystallites whose heights are al-

ways an integer number of Pb共111兲 interplanar distances
共2.86 Å兲. In large terraces of the substrate every Pb island
has a unique height, while in regions of the sample with a
high density of steps 共i.e., in step bunches such as the one
shown in Fig. 1兲, the islands may extend over several ter-
races of the substrate with their thickness changing by 1 ML
from terrace to terrace. This shape has been termed an
‘‘atomic wedge.’’14 The substrate steps crossed by the Pb
islands can be recognized at the island surface due to the
difference in the interplanar distance for lead 共2.86 Å兲 and
copper 共2.1 Å兲 in the 关111兴 direction. This difference pro-
duces a small step of 0.76 Å at the island surface that allows
us to unambiguously assign the thickness of Pb in every
portion of each island.
The statistical distribution of island heights in a sample
prepared with a given deposition rate and substrate tempera-
ture can be obtained through the analysis of several STM
images recorded on different areas of the sample. The size of
the images has been chosen to be 3000⫻3000 Å2 , that is,
large enough to ensure that they show a statistically signifi-
cant part of the film, while still allowing a clear identification
of the island heights. It is also important to determine the
possible influence of substrate steps in the observed height
distribution. To this end four examples of the STM images,
selected among those used to determine the island height
distribution of one and the same sample, are displayed in
Fig. 2. The images correspond to regions of the sample with
different step density. In regions with less steps there are less
islands, but they are larger. The local coverage varies slightly
over the sample surface due to inhomogeneities during FIG. 2. Four representative STM images obtained in the same
evaporation, but it does not depend on the step density. The sample but in areas with a different local slope. The images are
heights of the Pb islands in atomic layers are also indicated 3000⫻3000 Å2 , and a fraction of the differentiated image has been
in the figure. A casual inspection of the images reveals that added to enhance the visibility of the substrate and island steps.

115401-2
OBSERVATION OF PREFERRED HEIGHTS IN Pb . . . PHYSICAL REVIEW B 66, 115401 共2002兲

FIG. 4. Island height distributions 共percentage of area covered

FIG. 3. Island height distributions 共number of Pb islands with a by a given height兲 for Pb deposited on Cu共111兲 at 300 K and an-
given height兲 for different thicknesses of Pb deposited on Cu共111兲 nealed to 400 K. The data shown correspond to deposited nominal
at 300 K and annealed to 400 K. The data shown correspond to coverages of 2.5, 5, 7.5, and 16 ML, respectively. Regardless of the
deposited nominal coverages of 2.5, 5, 7.5, and 16 ML, respec- coverage deposited, some island heights are strongly preferred,
tively. while others are almost never observed. The peaks corresponding to
the ‘‘magically’’ preferred island heights are clearly resolved.

some island heights seem to appear more frequently than ing the thickness range from 2 to 20 equivalent ML. The
others, regardless the local slope of the substrate. For in- existence of the same ‘‘magic’’ heights preferred for Pb is-
stance, in the images of Fig. 2 several islands of 6, 8, 10–11, lands in the equilibrium distribution was confirmed for films
and 15 ML can be found, but not one appears with a height prepared with all different thicknesses.
of 5, 9, 12, or 16 ML. It is evident that the detailed height
distribution is different depending on the local concentration IV. LOCAL DETECTION OF QWS’S BY SCANNING-
of steps, but in all cases only certain heights are observed. TUNNELING SPECTROSCOPY
To quantify this observation we counted the number of
times that any given height appears on the surface without In the following we will demonstrate that the Pb islands
taking into account the size of the corresponding Pb island. behave as one-dimensional quantum wells along the 关111兴
In Fig. 3 we present the island height distribution for four direction, where the electronic states are quantized. The ef-
films with increasing nominal coverage of Pb. In all cases fective depths of the wells is given by the local height of the
some island heights appear with higher probability than oth- island. The lateral dimensions of the islands 共or even the
ers. For instance, there is a large number of islands with 6, 8, terraces within a given island兲 are too large to produce lateral
11, 15, or 17 atomic heights, while those with 7, 9, or 12 are confinement. The s-p electrons of Pb around the Fermi en-
much less abundant. In fact the less frequent 共‘‘forbidden’’兲 ergy are completely confined within the islands in the per-
heights are only observed on narrow terraces corresponding pendicular 共111兲 direction, by the vacuum barrier 共Pb work
to Pb islands that cross several substrate steps, sandwiched function兲 and the Cu bulk band gap, respectively. The con-
between the most commonly observed 共‘‘magic’’兲 heights. finement induces the discretization of the Pb sp band and the
It seems that the presence of substrate steps forces the corresponding QWS’s can be detected locally by scanning-
appearance of certain island heights that in a perfectly flat tunneling spectroscopy.14,15 Figure 5共a兲 shows representative
surface will be strictly forbidden. The existence of preferred dI/dV-V curves recorded at 300 K with the STM tip placed
or ‘‘magic’’ island heights is clearly revealed by weighting on top of Pb islands of varying height. The peaks in a given
the island distribution with the area covered by every island spectrum correspond to the different QWS’s associated to the
height. In Fig. 4 we show the percentage of area covered by thickness of the Pb island. The position in energy of the
islands of any given height for increasing thicknesses of Pb. QWS’s do not depend on the lateral size of the island, only
It is clear that certain thicknesses 共5, 9, 12, 13, 14, and on its height. The QWS spectra are independent of the tip
18 –19 ML兲 are forbidden, while 6, 8, 11, 15, 17, and 20 ML sample surface in the range used. The tick marks reflect the
are strongly preferred. We have deposited films of Pb cover- calculated energy positions of the QWS’s, accurately pre-

115401-3
OTERO, VÁZQUEZ DE PARGA, AND MIRANDA PHYSICAL REVIEW B 66, 115401 共2002兲

FIG. 6. Cut of the Fermi surface of Pb in the 共111兲 plane 共light

gray兲. The three nesting vectors along the 共111兲 direction are shown
as thick arrows. The upper diagram shows schematically that under
elastic reflection at the interfaces the perpendicular momentum
transfer must link two points of the constant energy surface. Due to
the confinement, the perpendicular momentum transfer is quantized.
V. ENERGETICS OF THE QWS’S AND THE NESTING
FIG. 5. 共a兲 Set of measurements of tunneling conductance ver- VECTORS OF THE FERMI SURFACE
sus sample voltage recorded at constant height above selected Pb The total energy of the Pb islands contains a term given
islands deposited on Cu共111兲. Notice the evolution of the quantized by the electronic energy of the occupied QWS’s. This con-
energy levels with the island height. The tick marks indicate the
tribution could outweigh the elastic energy and, eventually,
QWS’s calculated for an infinite quantum well of the given thick-
dictate the size distribution of the islands. The electronic
ness; 共b兲 spectroscopic image 700⫻700 Å2 of a Pb island contain-
energy of the occupied QWS’s will be maximized whenever
ing several quantum wells. The dc voltage is tuned to a QWS of the
central stripe 共1.15 V兲. This image was acquired by adding an ac
a QWS is placed right at the Fermi energy and it will de-
signal of 60 mV and 1.1 kHz. crease as the QWS passes above the Fermi level. Accord-
ingly, the ‘‘magic’’ heights would be those of islands which
dicted down to 4-ML-high islands by a simple, one- present a QWS well separated from the Fermi level and the
dimensional, infinitely high potential well model.15 The en- ‘‘forbidden’’ heights of islands with a QWS close to the
ergy levels can also be described with more realistic phase Fermi energy. In order to qualify this suggestion, it is neces-
accumulation models, taking into account the scattering sary to predict the actual heights of the Pb islands that will
properties of the Pb/Cu and the Pb/vacuum interfaces, but in present a QWS located at the Fermi level. This can be done
this case the position of the scattering planes must be slightly by means of the nesting vectors of the Fermi surface of Pb.16
shifted with respect to the interfaces, as has been discussed The confinement of the electrons inside the islands quantizes
elsewhere.15 their perpendicular momentum to a series of discrete values
Figure 5共b兲 shows a spectroscopic image of the empty given by k n ⫽n ␲ /Nd, with N being the number of Pb layers,
states recorded on top of one of the islands covering several d the interlayer distance, and n the 共integer兲 quantum
terraces of the substrate. The image has been taken with a number.15 The DOS at the Fermi level and the total elec-
bias voltage tuned to a QWS of the central stripe. That well, tronic energy will present a maximum due to the presence of
thus, appears bright, while those adjacent with a thickness a QWS whenever an n value satisfies k F ⫽k n .
corresponding to 1 ML more or less appear darker since in In Fig. 6 we show a cut of the calculated17 Fermi surface
the chosen energy window no QWS’s exist. The image of bulk Pb along 关111兴 displaying the nesting vectors and a
proves that, even when the islands extend over several ter- schematic diagram showing that q nesting coincides with the
races of the substrate and, then, present several atomic thick- first Brillouin-zone representation of 2k F,⬜ . The hole sur-
nesses, each island can be considered as an array of indepen- face in the second zone is indicated by the solid gray line,
dent quantum wells as long as the lateral size of each terrace while the dashed gray line indicates the electron pockets in
is larger than the Fermi wavelength of Pb. The large differ- the third zone. Then, the island heights 共Nd兲 with a maxi-
ence in energy between the QWS’s of wells whose thick- mum in the DOS at the Fermi level will be given by
nesses differ by one layer 共e.g., 0.21, 1.15, and 2.1 eV versus
⫺0.38, 0.65, and 1.65 eV兲 results in no mixing of states even 2␲
N⫽ n. 共1兲
in the same island. q nesting d

115401-4
OBSERVATION OF PREFERRED HEIGHTS IN Pb . . . PHYSICAL REVIEW B 66, 115401 共2002兲

Fermi level will present a maximum due to a QWS. We see

that for N⫽6, 8, 10, 11, 15, 17, and 20 layers, there is no
quantum number that yields a QWS close to the Fermi level.
These are precisely the ‘‘magic’’ heights permitted for the Pb
islands, as shown in the average distribution of island heights
depicted in the right panel of Fig. 7. On the contrary, there
are almost no islands with 5, 9, 12, 14, 18, 19, and 21 layers,
where one or more QWS’s are predicted to be present at the
Fermi energy. The heights most frequently observed are those
where the QWS’s are not close to the Fermi level, while for
every thickness strongly suppressed, at least one of the nest-
ing vectors satisfies Eq. (1), i.e., a quantum well state would
be located at E F . This suggests that the system structurally
rearranges itself to minimize the total energy, avoiding the
islands with these numbers of layers.

VI. CONCLUSIONS

In conclusion, we have observed that certain island

heights on films of Pb deposited on a Cu共111兲 surface are
strongly preferred while some others are almost completely
suppressed. The local, and simultaneously recorded topo-
graphic and spectroscopic data allow us to conclude that the
Pb nanocrystallites rearrange their height to avoid the ener-
FIG. 7. 共a兲 Left panel: The triangles indicate the Pb island
getically unfavorable situation of having a subband of QWS
heights, in number of layers, N, which would have a QWS at the and, thus, a high DOS, at the Fermi level. The energy con-
Fermi energy according to Eq. 共1兲. The heights are calculated for tribution of the QWS influences the resulting island mor-
the different quantum numbers n and the two relevant nesting vec- phology. The QWS’s with parallel components of the mo-
tors of the bulk Fermi surface of Pb. 共b兲 Right panel: Average of the mentum close to zero are directly detected by scanning-
distributions for different coverages. Notice that the ‘‘magic’’ island tunneling spectroscopy, while the others are deduced from
heights correspond to those that do not present an occupied QWS the nesting vectors of the Pb Fermi surface. In all the range
close to the Fermi energy. On the other hand, whenever the pre- explored the ‘‘magic’’ heights can be accurately predicted
dicted values 共triangles兲 come close to an integer value of Pb layers, from the nesting vectors of the Fermi surface of Pb. This
the corresponding island is strongly suppressed in the observed dis- constitutes the first direct experimental evidence of the influ-
tribution. ence of QWS’s in determining the equilibrium size of nanoc-
rystals. This experimental example of stable, ‘‘magic’’
The magnitude of these three nesting vectors is 1.23, 1.27, heights in nanoislands 共and possibly films兲 controlled by
and 1.79 Å⫺1 , respectively. The nesting vector electronic states originated by the 1D confinement, namely,
q 1 (1.23 Å⫺1 ) at k储 close to zero corresponds to the QWS the quantum well states, extends previous notions on the role
detected by tunneling spectroscopy. These kind of QWS’s of the confined motion of electrons in determining ‘‘magic’’
could also be detected by angle-resolved ultraviolet photo- sizes in nanoclusters3 共3D confinement兲 and nanowires20 共2D
emission spectroscopy 共ARUPS兲 at normal emission.18 confinement兲. In both cases, geometric structures minimizing
The nesting vectors at nonzero parallel momentum, the energies of the occupied states relative to the Fermi level
q 2 (1.27 Å⫺1 ) and q 3 ⫽1.71 Å⫺1 , give rise to QWS not seem to be preferred. This ‘‘electronic-controlled’’ mode of
seen by tunneling spectroscopy, but detectable by ARUPS growth could have implications for the future wave-function
with k 储 different from zero.19 The nesting vector q 2 will give engineering design of stable nanometer-scale materials and
almost the same periodicity in the DOS at the Fermi level devices.
and in the total electronic energy as q 1 , while the nesting
vector q 3 will contribute little to the electronic energy due to ACKNOWLEDGMENTS
the curvature of the Fermi surface at the connected points.
In Fig. 7 we plot the values of N corresponding to Eq. 共1兲 This work has been financed by the Dirección General de
for different values of the quantum number, n, and for the Investigación, Ciencia y Tecnologı́a 共DGICyT兲 with Grant
two nesting vectors that contribute more to the total elec- No. BFM2001-0174 and by the Consejerı́a de Educación
tronic energy. Whenever N becomes an integer for some from the Comunidad de Madrid with Grant No. 07N/0055/
value of the quantum number, n, the density of states at the 2001.

115401-5
OTERO, VÁZQUEZ DE PARGA, AND MIRANDA PHYSICAL REVIEW B 66, 115401 共2002兲

*Present address: CAMP and Department of Physics and As- 11

A.L. Vázquez de Parga, O. S. Herman, R. Miranda, A. Levy
tronomy, University of Aarhus, DK-8000 Aarhus C, Denmark. Yeyati, N. Mingo, A. Martin Rodero, and F. Flores, Phys. Rev.
Electronic address: [email protected] Lett. 80, 357 共1998兲.
1 12
A. Cebollada, R. Miranda, C.M. Schneider, P. Schuster, and J. This is in opposition to the nearly layer-by-layer growth mode of
Kirschner, J. Magn. Magn. Mater. 102, 25 共1991兲. the same system at lower temperatures described by B.J. Hinch,
2
J.E. Ortega and F.J. Himpsel, Phys. Rev. Lett. 69, 844 共1992兲. C. Koziol, J. P. Toennies, and G. Zhang, Europhys. Lett. 10, 341
3
W.D. Knight, K. Clemenger, W.A. de Heer, W.A. Saunders, M.Y. 共1989兲.
Chou, and M.L. Cohen, Phys. Rev. Lett. 52, 2141 共1984兲. 13
J. Camarero, J. Ferren, V. Cros, L. Gomez, A.L. Vazquez de
4
A.R. Smith, K.-J. Chao, Q. Niu, and C.-K. Shih, Science 273, 226 Parga, J.M. Gallego, J.E. Prieto, J.J. de Miguel, and R. Miranda,
共1996兲. Phys. Rev. Lett. 81, 850 共1998兲.
5
L. Gavioli, K.R. Kimberlin, M.C. Tringides, J.F. Wendelken, and 14
I.B. Altfeder, K.A. Matveev, and D.M. Chen, Phys. Rev. Lett. 78,
Z. Zhang, Phys. Rev. Lett. 82, 129 共1999兲.
6 2815 共1997兲.
L. Huang, S.J. Hay, and J.H. Weaver, Surf. Sci. 416, L1101 15
R. Otero, A.L. Vázquez de Parga, and R. Miranda, Surf. Sci. 447,
共1998兲.
7 143 共2000兲.
V. Yeh, L. Berbil-Bautista, C.Z. Wang, K.M. Ho, and M.C. Tring-
ides, Phys. Rev. Lett. 85, 5158 共2000兲.
16
M.D. Stiles, Phys. Rev. B 48, 7238 共1993兲.
8
W.B. Su, S.H. Chang, W.B. Jian, C.S. Chang, L.J. Chen, and T.T.
17
J.R. Anderson and J.A.V. Gold, Phys. Rev. 139, 1459 共1965兲.
Tsong, Phys. Rev. Lett. 86, 5116 共2001兲.
18
T.-C. Chiang, Surf. Sci. Rep. 39, 181 共2000兲.
19
9
Z. Zhang, Q. Niu, and C.-K. Shih, Phys. Rev. Lett. 80, 5381 P. Segovia, E.G. Michel, and J.E. Ortega, Phys. Rev. Lett. 77,
共1998兲. 3455 共1996兲.
10
D.A. Luh, T. Miller, J.J. Paggel, M.Y. Chou, and T.-C. Chiang,
20
A.I. Yanson, I.K. Yanson, and J.M. van Ruitenbeek, Nature 共Lon-
Science 292, 1131 共2001兲. don兲 400, 144 共1999兲.

115401-6