2.1 The structure of medium sized silicon cluster anions
J. Müller and G. Ganteför in collaboration with B. Liu and K.-M. Ho
(Ames Laboratory and Department of Physics & Astronomy, Iowa State University, Ames, Iowa 50011, USA) S. Ogut and J.R. Chelikowsky
(Department of Chemical Engineering & Materials Science, University of Minnesota, Minneapolis, Minnesota 55455, USA)
A.A. Shvartsburg and K. W. M. Siu
(Department of Chemistry, York University, Toronto, Ontario, Canada M3J 1P3) A large number of substances had been thought to
displace silicon as the most important semiconductor material, but still none can compete with it as of today.
Instead, the capabilities of semiconductor devices have been increased via their continuous miniaturization. This has induced a growing interest in the properties of small Si structures, including the clusters containing just a few atoms.
Fig.1: Photoelectron spectra for Sin anions (n = 8-20).
Experimental scans are in thin lines, thick lines are for the DFT-LDA simulations.
Little had been known for sure about the geometries and electronic structure of medium-sized Si clusters in the 10-50 atom range. Ion mobility measurements are often too blunt as a structural probe because different isomers often have very similar mobilities 1). The other approach involves a comparison of measured electronic
properties such as electron affinities with simulations for candidate geometries. For some time, the structural op-timization of Si clusters had stalled at about the size of 10 atoms. At this point the number of possible isomers becomes too large to be treated using the then known methods: exhaustive permutation of all options by man-ual construction and simulated annealing.
A powerful tool to determine the electronic structure is Photoelectron Spectroscopy (PES) 2) that reveals the energies of molecular orbitals. In a PES experiment, size-selected cluster anions are irradiated using a fixed-wavelength laser producing photons with energy above the detachment threshold of investigated species. The distribution of kinetic energies of released electrons is measured. These data reveal the electronic structure of neutral clusters. Previously published information for the PES of Sin anions has been limited to n £ 13 because of insufficiently high signal intensities for larger sizes.
Here we report a major progress in the structural characterization of medium-sized Si clusters enabled by advances in both experiment and theory. On the side of experiment, we have obtained the photoelectron spectra for Sin anions with up to 50 atoms. This has been made possible by a specially tuned Pulsed Arc Cluster Ion Source (PACIS) 3) that produces a substantially brighter ion beam than those available previously. The source features a 30 cm long extender with diameter of 4 mm, including a 4 cm long waiting room 8 mm in diameter.
Liquid nitrogen circulating through this extender effi-ciently cools the clusters produced. This has allowed us to resolve multiple features in the PES, thus determining the vertical detachment energies and HOMO-LUMO gaps for medium-sized clusters. On the side of theory, a novel tool for molecular optimization, the genetic algo-rithm 4), has recently been applied to Sin species in the n £ 20 size range 5). A superior power of this method has allowed us to proceed beyond the previous limit of about 10 atoms. Specifically, we have accumulated an exten-sive set of low-energy isomers for Sin anions with n £ 20. All these have been optimized using the density functional theory (DFT) in both local density and gener-alized gradient approximations. Photoelectron spectra for candidate geometries have been simulated using DFT-LDA. A comparison of results with the
measure-ments has allowed us to make the structural assignments.
Fig. 2: Vertical detachment energies: measured values (solid line), calculated for the Sin anion ground states (filled circles), calculated for other low-energy Sin- iso-mers (other symbols).
For the purpose of initial screening, we had calcu-lated the vertical detachment energies (VDEs) for a number of low-energy Sin anion isomers. These are compared with the measurements in Fig. 2. By inspec-tion, the VDEs for Sin anion ground states match the measured values for all n £ 20, while other low-energy isomers fail to reproduce the measurements (with a few exceptions limited to n = 11 and 12). For species exhib-iting a reasonable agreement, the complete PES was simulated and compared with experiment. The results for n = 8 - 20 are presented in Fig. 1. Simulations for the lowest-energy geometries found by genetic algorithm are superimposed (the data for three near-degenerate isomers competing for the global minimum are shown for n = 11). The agreement between calculations and experiment is excellent except for n = 12. Since a photo-electron spectrum is highly specific to cluster structure, there is a high degree of confidence in that the lowest-energy geometries found in our calculations are actually observed in experiment. A more detailed discussion can be found in Ref. 5).
Fig. 3: HOMO-LUMO gap calculated for Sin (n £ 20) species in the geometries that are lowest-energy for an-ion (solid line) and neutral (dashed line). Symbols mark the measurements for the case of anions.
The evolution of band gap (HOMO-LUMO gap) in Si clusters as a function of size is of great interest from both fundamental and applied viewpoints. These gaps can be extracted from the PES or calculated. For metal-lic species with delocalized electrons, the gap in finite particles tends to decrease with increasing cluster size.
However, the gap for Sin (n £ 20) neutrals varies ir-regularly in the 1 - 2 eV range without apparent decrease for larger clusters (Fig. 3).
In conclusion, we have elucidated the structure of medium-sized Si clusters using photoelectron spectros-copy. These species are basically built from the Si9 tri-capped trigonal prism (TTP) subunits 6), although some other recurrent structural patterns become important for n approaching 20. The dissociation energies of Si clus-ters drastically decrease for n > 10 7), so it is hardly sur-prising that the ion yield drops dramatically. Further-more, photoelectron spectra of larger Si clusters display peculiar features for some sizes, including n = 33 and 43. This may either reveal a HOMO-LUMO gap sub-stantially greater than that for other sizes, or indicate an unusual predominance of one geometry in the isomeric mixture. This behavior is not yet understood and is a subject of our ongoing research.
(1) R.R. Hudgins et al., J. Chem. Phys. 111 (1999) 7865 (2) H. Handschuh et al., Rev. Sci. Instrum. 66 (1995) 3838 (3) Chia-Yen Cha et al, Rev. .Sci. Instrum. 63 (1992) 5661 (4) K.M. Ho et al, Nature 392 (1982) 582
(5) A.A. Shvartsburg et al., J. Chem. Phys. 112 (2000) 4517 (6) J. Müller et al, submitted to Phys Rev. Lett.
(7) A.A. Shvartsburg et al., Phys. Rev. Lett. 81 (1998) 4616
4 6 8 10 12 14 16 18 20
0,0 0,5 1,0 1,5 2,0
Bandgap [eV]
Number of Atoms
2.2 Time resolved dynamics of electronic excitations in C
3-S. Minemoto, J. Müller, R. Fromherz, G. Ganteför, H.J. Münzer, J. Boneberg and P. Leiderer The development of femtosecond lasers has made it
possible to study fast dynamical processes in atoms, molecules and condensed matter. A further step forward was the combination of femtosecond lasers with pho-toelectron spectroscopy, which allowed the direct obser-vation of the reorganization of the electronic structure after photoexcitation. In such an experiment, the pump pulse triggers dynamical processes like single particle excitations or fragmentation, and with the UV-probe pulse a photoelectron spectrum is recorded at a given delay. The series of photoelectron spectra reveals the time evolution of the system with increasing pump / probe delay. Here we describe the application of this technique to study the decay of electronic excita-tions in mass selected nanoparticles and clusters, where the time scales and decay mechanisms might be differ-ent from the bulk properties as a result of the finite size of the particles.
For small aggregates with a well-defined number of atoms (clusters) it is known that the properties may vary with each additional atom. Therefore, for experiments on clusters mass separation is essential. One successful method used for the study of the ground state electronic structure of clusters is photoelectron spectroscopy of negatively charged ions 1). Recently, this technique has been combined with femtosecond lasers to study fast fragmentation processes in molecules and clusters like I2- 2) and Au3- 3).
In the present article, we present the application of this technique to observe the time-evolution of elec-tronic excitations in mass selected clusters. As a first example, we have studied the decay of an excited state of the carbon trimer anion C3-. Taking advantage of the new technique we determined the lifetime of this excited state directly. In addition, the photoelectron spectra re-veal the nature of the participating electronic states un-ambiguously. Especially, it was possible to directly termine the neutral "parent state" (see below) by de-tachment from the excited resonance proving the as-signment to a Feshbach resonance. This example de-monstrates the power of the method for studying elec-tronic excitations in larger clusters and nanoparticles.
Details of the experimental set up have been de-scribed elsewhere 1,3). The anions are generated directly in a pulsed arc cluster ion source (PACIS) and mass separated with a time-of-flight mass spectrometer. A selected bunch of anions is irradiated by the pump and probe laser pulses and the kinetic energy of the detached electrons is measured using a "magnetic-bottle"-type time-of-flight electron spectrometer. The femtosecond laser is split into pump and probe pulse, which both have equal intensities (~ 1mJ / cm2) and pulse widths (~ 300 fs).
Fig. 1 displays a comparison of two photoelectron spectra of C3- obtained with zero delay (trace 1) and
with large delay (13 ps, trace 2) between pump and probe pulse. Both spectra have been obtained with a photon energy of hn = 3.1 eV. The spectra are normal-ized to the intensity of peak A located at a kinetic energy of 1.1 eV. This corresponds to the difference of the photon energy and the electron affinity of C3 -(EA = 1.995 eV), and peak A is assigned to direct de-tachment from the electronic ground state of C3-. The final state is the ground state of neutral C3.
Fig. 1: Pump / probe photoelectron spectra of C3-. Feature B is located at a kinetic energy of 2.0 eV. Its intensity with respect to the main feature A varies de-pending on the time delay between the two laser pulses.
This indicates that its origin might be due to a two pho-ton process. As a first idea one would expect an increase of hn = 3.1 eV of the kinetic energy (corresponding to a kinetic energy of 4.3 eV) of an electron detached in a 2-photon process. Such a peak is observed too, but with an extremely low intensity (peak C in Fig. 1) and will be discussed later. For the 2-photon feature B an increase of 0.9 ± 0.1 eV only is observed and there are 2.2 eV of energy missing.
The photoelectron data can be understood if we as-sume the existence of an excited state of the anion lo-cated at an energy above the electron affinity. There are two different types - shape resonances and Feshbach resonances. For shape resonances, the electron is weakly bound in the potential of the neutral core by an angular momentum barrier and the decay occurs by tunneling through the barrier. The "parent state" is the state of the neutral atom (ground or excited state) and the decay of the shape resonance into this neutral state is energeti-cally allowed. For Feshbach resonances, the electron is bound with a positive electron affinity to the neutral core, which must be in an excited electronic state (the
"parent state" of the Feshbach resonance). Autodetach-ment occurs via electronic autodetachAutodetach-ment, i.e., the neutral core relaxes into its ground state transferring the energy to the additional electron which is ejected. In contrast to shape resonances, the decay occurs by
con-0 1 2 3 4 5
certed action of at least two electrons resulting in con-siderably longer lifetimes.
If, in case of a Feshbach resonance, the excited anion is hit by a second photon, the additional electron is re-moved leaving the neutral in the excited "parent state" of the resonance. The kinetic energy of such an electron is increased with respect to the electron from direct photodetachment by the energy of the second photon (hn = 3.1 eV) minus the excitation energy of the neutral
"parent state". Indeed, there is a known excited state of neutral C3 with the proper energy: the first excited state with an energy of 2.152 eV. We conclude that at zero delay electrons are detached from a Feshbach resonance of C3-.
Fig. 2: Autodetachment (schematic). Three mechanisms are shown for the process of detaching an electron, to-gether with the corresponding electronic states.
There are at least three processes induced by the in-teraction with the laser beam (Fig. 2):
I. direct detachment from the ground state (1-photon, peak A):
C3- + hn à C3 + e- (Ekin = 1.0 eV) II. resonant autodetachment (1-photon, peak A):
C3- + hn à C3-* à C3 + e- (Ekin= 1.0 eV) III. direct detachment from the excited state (2-photon,
peak B):
C3- + hnpump à C3-*
C3-* + hnprobe à C3* + e- (Ekin = 1.9 eV).
The kinetic energies of the detached electrons are given for photon energies of hn=hnpump=hnprobe=3.1 eV.
Processes I and II yield identical features in the electron spectra (peak A in Fig. 1), because initial and final state are the same for both 1-photon channels. The 2-photon process III results in the appearance of feature B.
Feature C in Fig. 1 can be explained by a fourth process:
IV. "shake down" detachment from the excited state (2 photon, peak C):
C3- + hnpump à C3-*
C3-*+ hnprobe à C3 + e- (Ekin = 4.2 eV).
The difference between processes III and IV is the fi-nal state of the neutral C3, which is the ground state in case of process IV.
Fig. 3: Relative intensity of peak B (Fig. 1) as a function of the pump/probe delay.
In a pump / probe experiment, the lifetime of an ex-cited state can be measured directly. Fig. 3 displays the dependence of the intensity of peak B in Fig. 1 on the delay between the pump and the probe laser pulses. Ac-cording to process III, this dependence is directly related to the lifetime of the Feshbach resonance (noted C3-*).
From an exponential fit (Fig. 3) the lifetime t of the Feshbach resonance of C3-is determined to be
t = 2.6±0.7ps.
In conclusion, we present results of the application of time-resolved photoelectron spectroscopy to study the decay of an excited electronic state in mass selected cluster anions. As a first example, we studied the elec-tronic autodetaching process of a Feshbach resonance of C3- and could distinguish four different channels con-tributing to the photoelectron signal. The use of charged particles allows for an accurate mass separation and in the future the method will be applied to the study of various electronic excitations in clusters and nanoparti-cles.
(1) H. Handschuh, G. Ganteför and W. Eberhardt, Rev. Sci.
Instrum. 66 (1995) 3838
(2) B.J. Greenblatt, M.T. Zanni, D.M and Neumark, Chem.
Phys. Lett. 258 (1996) 523
(3) G. Ganteför, S. Kraus and W. Eberhardt, J. Electr.
Spectr. Rel. Phen. 35 (1997) 88
vacuum
2.3 Deposition of mass selected aluminum clusters
B. Klipp, M. Grass, U. Lutz, G. Ganteför,
T. Schlenker, J. Zimmermann, J. Boneberg and P. Leiderer Deposition of mass selected clusters is a new method
for the preparation of well defined nanostructures on surfaces. The preparation of nanostructures on surfaces is one of the major tasks in technology and basic re-search. Applications of nanostructured surfaces are abundant and cover a wide range from heterogeneous catalysis to high density computer memories. From the point of view of basic research, many of the properties of small 3-dimensional nanostructures and 2-dimen-sional islands on surfaces are not well understood yet.
E.g., there is no systematic study of the size-dependence of the electronic structure of clusters of simple metals on surfaces, although free clusters of such metals show strong variations depending on the number of delocal-ized electrons.
In the last annual report we described a new experi-mental set up including a cluster source which is oper-ated at high repetition rates (up to 1000 Hz), ion extrac-tion, mass selection using a 45° sector magnet and an ion optics to allow soft landing of clusters on a surface.
With our experimental set up currents of monodispersed cluster ions of 1012 atoms per second can be achieved.
The width of the kinetic energy distribution is in the range of 1 eV which is necessary to decelerate the clus-ter ion beam (“soft landing”).
Fig 1: XPS on deposited aluminum monomers.
To minimize the interaction between the substrate and the clusters we chose a graphite (HOPG = highly orien-tated pyrolytic graphite) substrate, which is an inert Van der Waals surface. In the first part of the experiment we deposit the aluminum monomer on HOPG. The amount of aluminum is 2·1014 atoms which corresponds to 10 % of a monolayer and the kinetic energy is varied from 5 eV, 10 eV to 40 eV.
The photoelectron spectra show a dramatic change between the 5 eV and the 10 eV deposition on the one hand and the 40 eV deposition on the other hand (Fig. 1). The 5 eV and 10 eV samples exhibit two peaks in the Al 2p spectra, one peak corresponding to Al-Al bonds and one corresponding to reacted Al. The Al-Al bonds indicate the formation of aluminum islands.
Fig. 2: STM-Picture: Al1 / HOPG, 5 eV deposition.
Fig. 3: STM-Picture: Al1 / HOPG, 10 eV deposition.
Deposition of aluminum monomers with a kinetic en-ergy of 40 eV, which is a few times the binding enen-ergy
82 80 78 76 74 72 70 68 66 64 62
43 % Al-Al 57 % reacted Al 2p
Intensity [a. u.]
Binding Energy [eV]
E
KIN= 10 eV
82 80 78 76 74 72 70 68 66 64 62
100 % reacted Al 2p
Intensity [a. u.]
Bi nding Energy [e V]
E
KIN= 40 eV E
KIN= 5 eV
82 80 78 76 74 72 70 68 66 64 62
46 % Al-Al 54 % reacted Al 2p
Intensity [a. u.]
Binding Energy [eV]
of a C atom in the graphite bulk, causes sputtering of the surface and implantation of single atoms. There are no Al-Al bonds, and the amount of oxygen (not shown) is five times higher, because the sputtered surface is not inert.
The STM pictures support the interpretation of the XPS spectra. Deposition with kinetic energies below the binding energy of a C atom in graphite (e.g. 5 eV) is comparable to thermal vapor deposition. The STM pic-ture shows islands with more than 1000 atoms, which can be moved by the STM tip (Fig. 2). Deposition with kinetic energies in the range of the binding energy of a C atom in graphite (10 eV) leads to smaller islands (about 150 atoms). Few atoms with the highest energy are able to cause a defect in the surface. These defects act as nucleation sites for island growth and the islands are pinned (Fig. 3). The 40 eV deposition shows no islands at all.
Following that we deposit 2.8·1012 Al70±3 clusters on HOPG, which is the same amount of atoms as in the first part (2·1014 atoms, 10 % of a monolayer). The kinetic energies are 10 eV, 40 eV and 80 eV.
Fig. 4: XPS on deposited Al70±3 clusters.
Non of the spectra show evidence for formation of bigger islands, the Al-Al peak is smaller than the reacted Al peak. This means the islands are in the size of 70 atoms, the size of the clusters. The only difference in the 3 spectra is the bigger Al-Al peak in the 10 eV spectra (Fig. 4). This may be a hint for the formation of a few bigger island, consisting of two or three clusters or it may be, that these clusters are not pinned at the surface.
If the clusters are pinned, the Al atoms on the substrate are reacted (Al-C), if the clusters are just “lying on the surface” the atoms on the surface are not reacted (Al-Al). In both cases these clusters are soft landed. STM investigations of a heated 10 eV deposition show clus-ters at surface steps, which also is an indication for mo-bile, i.e. soft landed clusters.
Fig. 5: STM picture: Al70 / HOPG , 80 eV.
The deposition of aluminum clusters consisting of 70±3 atoms with kinetic energies of 40 eV or more re-sults in single nanostructures pinned on the surface, which can be imaged by a STM with atomic resolution (Fig. 5).
82 80 78 76 74 72 70 68 66 64 62
69 % reacted 31 % Al-Al Al 2p
Intensity [a. u.]
Binding Energy [eV]
82 80 78 76 74 72 70 68 66 64 62
86 % reacted 14 % Al-Al Al 2p
Intensity [a. u.]
Binding Energy [eV]
E
KIN= 10 eV
82 80 78 76 74 72 70 68 66 64 62
13 % Al-Al 87 % reacted Al 2p
Intensity [a. u.]
Binding Energy [eV]
E
KIN= 40 eV
E
KIN= 80 eV
2.4 A new experimental setup for the in-situ investigation of the electronic, vibrational and chemical properties of monodisperse supported clusters
J. Müller, S. Burkart, R. Fromherz, C. Peucker, S. Boisson and G. Ganteför Clusters deposited on surfaces have moved into the
focus of interest in surface physics as a novel method to prepare nanostructured surfaces with exceptional elec-tronical and chemical properties. This development is also due to the fact that any application of clusters be-yond the utilization of cluster beams requires deposition on a substrate.
We report a new experimental setup for the situ in-vestigation of small monodisperse supported clusters.
This instrument consists of a high intensity Pulsed Arc
This instrument consists of a high intensity Pulsed Arc