Shashank S Harivyasi1, Oliver T Hofmann1, Nahid Ilyas2, Oliver L A Monti2,3, Egbert Zojer1. 1. Institute of Solid State Physics, NAWI Graz, Graz University of Technology, Petersgasse 16, 8010 Graz, Austria. 2. Department of Chemistry and Biochemistry, University of Arizona, 1306 E. University Blvd., Tucson, Arizona 85721, United States. 3. Department of Physics, University of Arizona, 1118 E 4th Street, Tucson, Arizona 85721, United States.
Abstract
In this article, we investigate the interface between shuttlecock-shaped chloro boron-subphthalocyanine molecules and the Cu(111) surface. We highlight how molecular planarization induced by van der Waals forces can fundamentally alter the interface properties and how it can enable a particularly strong hybridization between molecular and metal states. In our simulations, we start from a situation in which we disregard van der Waals forces and then introduce them gradually by rescaling the interaction parameter, thereby "pulling" the molecule toward the surface. This reveals two adsorption regimes with significantly different adsorption distances, molecular conformations, and adsorbate-induced changes of the work function. Notably, the above-mentioned massive hybridization of electronic states, also observed in photoelectron spectroscopy, is obtained solely for one of the regimes. We show that this regime is accessible only as a consequence of the planarization of the molecular backbone resulting from the van der Waals attraction between the molecule and the surface. The results of this study indicate that for certain metal-molecule combinations unusually strong interfacial electronic interactions can be triggered by van der Waals forces creating a situation that differs from the usually described cases of physisorptive and chemisorptive interactions.
In this article, we investigate the interface between shuttlecock-shaped chloro boron-subphthalocyanine molecules and the Cu(111) surface. We highlight how molecular planarization induced by van der Waals forces can fundamentally alter the interface properties and how it can enable a particularly strong hybridization between molecular and metal states. In our simulations, we start from a situation in which we disregard van der Waals forces and then introduce them gradually by rescaling the interaction parameter, thereby "pulling" the molecule toward the surface. This reveals two adsorption regimes with significantly different adsorption distances, molecular conformations, and adsorbate-induced changes of the work function. Notably, the above-mentioned massive hybridization of electronic states, also observed in photoelectron spectroscopy, is obtained solely for one of the regimes. We show that this regime is accessible only as a consequence of the planarization of the molecular backbone resulting from the van der Waals attraction between the molecule and the surface. The results of this study indicate that for certain metal-molecule combinations unusually strong interfacial electronic interactions can be triggered by van der Waals forces creating a situation that differs from the usually described cases of physisorptive and chemisorptive interactions.
In
organic (opto)electronic devices, molecules are often in direct
contact with metal electrodes. This makes a comprehensive understanding
of metal–organic interfaces a prerequisite for improving device
performance. A key factor that dominates the properties of such interfaces
is the strength of the interaction between the metal and the molecules.
Together with the electronic structure of these interfaces, the interactions
have been explained by invoking a number of phenomena, such as exchange
interaction, dipole interaction, charge transfer, and covalent bonding.[1−10]The present study focuses on understanding peculiarities of
the
interfacial interactions and the electronic structure that may arise
from adsorption-induced geometry changes in certain nonplanar molecules.
In particular, we present a detailed computational study of the adsorption
of chloro boron-subphthalocyanine (ClB-SubPc) on Cu(111) and corroborate
it with an assessment of the interfacial electronic structure via
photoelectron spectroscopy. The system serves as a prototypical example
for the hitherto rarely discussed[11] scenario
in which the electronic structure of an interface is fundamentally
altered by changes of molecular conformation that are, in turn, enabled
by the van der Waals (vdW) forces between the molecule and the substrate.Beyond the fundamental character of the present study, it is also
motivated by the increasing technological relevance of subphthalocyanines
(SubPcs), which calls for an improved understanding of their properties.
SubPcs are optically active dyes that exhibit fluorescence and a high
second-order linear response.[12,13] They have been used
in organic photovoltaic cells,[14] in light-harvesting
systems,[15] and for blending-driven band
gap engineering.[16] Consequently, film formation
of the archetypical subphthalocyanine, ClB-SubPc, on different metal
surfaces has been the focus of numerous studies,[17−22] all revealing feature-rich thin films. The latter is, at least in
part, a consequence of the molecule’s peculiar geometry: A
ClB-SubPc molecule has a threefold rotational symmetry, combined with
a conical “shuttlecock-shaped” nonplanar geometry (Figure ).
Figure 1
Chemical structure of
chloro boron-subphthalocyanine (ClB-SubPc).
Upon adsorption, the 18 carbon atoms of the three benzene subunits
of the molecule (shaded gray here) are found closest to the Cu substrate.
They are collectively referred to as C-18 in this article.
Chemical structure of
chloro boron-subphthalocyanine (ClB-SubPc).
Upon adsorption, the 18 carbon atoms of the three benzene subunits
of the molecule (shaded gray here) are found closest to the Cu substrate.
They are collectively referred to as C-18 in this article.Consequently, depending on the molecular orientation
with respect
to the substrate upon adsorption, multiple plausible adsorption situations
can be envisaged.[22] This results, for example,
in charge transfer from the surface to the chlorine atom on Ag(111),[18] in the formation of bilayer nanocrystals on
Cu(111),[20] or in surface-catalyzed dechlorination
of (a fraction of) the molecules on Cu(111).[22] In the last case, one observes the formation of chemically different
ClB-SubPc domains consisting of either dechlorinated or intact molecules.[22] Here, we focus on molecules that remain intact
while adsorbing on Cu(111), while the most relevant data for dechlorinated
molecules are compiled in the Supporting Information (SI). The case of intact molecules is particularly insightful for
understanding the peculiarities that can arise from the interaction
between a nonplanar π-electron system and a coinage metal substrate:
By tuning the magnitude of van der Waals forces when simulating the
ClB-SubPc/Cu(111) interface, we are able to show how the van der Waals-induced
planarization of (significant parts of) the molecular backbone upon
adsorption gives rise to two distinct adsorption regimes. Moreover,
the planarization enables a particularly strong electronic coupling
between substrate and adsorbate. This, in turn, fundamentally changes
interface properties like the adsorbate-induced work function modification,
the local density of states, and the interfacial charge rearrangement.
Methodology
Computational
For our simulations,
we relied on density functional theory (DFT) as implemented in the
Fritz Haber Institute ab initio molecular simulations (FHI-aims) package,[23] employing the default “tight”
settings. We used the Perdew–Burke–Ernzerhof (PBE) functional[24] complemented by a surface-screening-corrected
van der Waals energy term,[25] a scheme commonly
designated as PBE + vdWsurf. Employing the repeated-slab
approach,[26] we studied the adsorption process
at a moderate packing density[22] with just
one molecule in every 7 × 4√3 orthogonal surface unit
cell. This represents the smallest orthogonal cell in which the energy
contribution from intermolecular interactions has decayed to less
than 0.1 eV. For simulating the isolated molecule in the gas phase,
we also used the FHI-aims package and employed open boundary conditions
with equivalent computational settings (the resulting coordinates
are provided in the SI).The Cu(111)
surface was represented by a slab with a thickness of six atomic layers,
of which only the top two were allowed to relax in the geometry optimizations.
In the direction perpendicular to the surface, a vacuum gap of at
least 25 Å was introduced together with a self-consistently determined
discontinuity in the electrostatic energy to decouple periodic replicas
of the slab.[27] The convergence criterion
for the self-consistent field cycles was set to 10–6 eV for the total energy, and the geometry was relaxed until the
maximum residual force component per atom was less than 0.01 eV/Å.
The states were broadened using a Gaussian function, and also a Gaussian
occupation scheme[28] was employed. For both,
σ was set to 0.2 eV. A 4 × 4 × 1 Γ-centered k-point grid was used for all periodic calculations, and
van der Waals corrections were not included for Cu–Cu atom
pairs in calculations containing the Cu slab. The lattice constant
for Cu, obtained with the above settings is 3.637 Å.
Experimental Section
ClB-SubPc was
purchased from Sigma-Aldrich and was purified by repeated gradient
sublimation in a custom-built furnace. It was degassed overnight in
a home-built Knudsen cell under ultrahigh vacuum conditions and at
temperatures slightly below the sublimation point. A polished Cu(111)
crystal of 99.999% purity (Princeton Scientific) was cleaned by repeated
cycles of Ar+ sputtering and annealing, whereas highly
oriented pyrolytic graphite (HOPG) was freshly cleaved and annealed
at 450 °C in vacuum. Molecular deposition occurred at room temperature
on both substrates, at a rate of 0.1 Å/min. Photoelectron spectroscopy
was carried out using a He(I) UV source (SPECS UVS 10/35) incident
at a 30° angle in a VG EscaLab MK II photoelectron spectrometer.
Spectra were acquired with an energy resolution of 89(8) meV.
Results and Discussion
Planarization and Hybridization
of ClB-SubPc
upon Adsorption on Cu(111)
The shuttlecock-shaped geometry
of the ClB-SubPc molecule (Figure a) is lost once it adsorbs on Cu(111). The molecule
then resembles a planarized disk with only a trigonal pyramidal feature
(composed of chlorine, boron, and three pyrrolic nitrogen atoms) preserved
in its center (Figure b). The 18 carbon atoms closest to the substrate (collectively termed
as C-18; see shaded rings in Figure ) end up lying approximately in the same plane, with
the standard deviation in their z-coordinates as low as 0.03 Å. As a more
accessible
quantification, in the following we define the planarity, Π,
of any given ClB-SubPc conformation via the standard deviation in
the heights of the C-18 atoms asThis definition implies that the molecule
in its gas-phase geometry has a planarity, Π, of 0%, whereas
Π is 100% when all C-18 atoms are in the same plane. For the
relaxed ClB-SubPc/Cu(111) interface, a planarity of 94% is obtained
for the adsorbed molecule. Concomitantly, the molecule’s “shuttlecock
angle”, the obtuse angle between the central symmetry axis
and any of the six peripheral carbon atoms in the benzene subunits,
decreases by 13° (from 116 to 103°, Figure a,b).
Figure 2
Structure of the ClB-SubPc molecule (a)
before and (b) after adsorption
on Cu(111). (c) Molecule-projected density of states (PDoS) of the
ClB-SubPc molecule adsorbed on the Cu(111) substrate (black curve)
and in the gas phase (red curve). The DoS of the gas-phase molecule
has been Gaussian broadened and shifted such that the vacuum level
of the molecule is the same as that of a clean Cu slab. The Fermi
level of that slab is then used as the energy reference for the isolated
molecule in the above plot. Panel (b) also indicates the unit cell
and the coordinate system used in all calculations employing periodic
boundary conditions. The coloring scheme is as follows: chlorine,
red; boron, green; nitrogen, blue; carbon, gray; hydrogen, yellow;
and copper, brown.
Structure of the ClB-SubPc molecule (a)
before and (b) after adsorption
on Cu(111). (c) Molecule-projected density of states (PDoS) of the
ClB-SubPc molecule adsorbed on the Cu(111) substrate (black curve)
and in the gas phase (red curve). The DoS of the gas-phase molecule
has been Gaussian broadened and shifted such that the vacuum level
of the molecule is the same as that of a clean Cu slab. The Fermi
level of that slab is then used as the energy reference for the isolated
molecule in the above plot. Panel (b) also indicates the unit cell
and the coordinate system used in all calculations employing periodic
boundary conditions. The coloring scheme is as follows: chlorine,
red; boron, green; nitrogen, blue; carbon, gray; hydrogen, yellow;
and copper, brown.Putting the planarization
of the ClB-SubPc π-system into
the context of similar molecules is difficult. With three instead
of four isoindole units and meso N atoms, boron-subphthalocyanines
are, so far, the only known lower homologues of phthalocyanine (Pc)
molecules. Although the adsorption of numerous nonplanar Pcs has been
studied extensively,[29−40] most have comparably smaller shuttlecock angles and thus do not
lend themselves to direct comparison. For instance, the nearest comparable
study of SnPc on Ag(111) also employing dispersion-corrected DFT reports
that adsorption results in a change of 5° (from 105 to 100°)
in the molecule’s shuttlecock angle when it adsorbs in a similar
orientation.[36] On the other extreme, there
are spherical- and bowl-shaped molecules like C60 and corannulene
(C20H10) that appear not to flatten on Cu(111),
and for which the overall geometry remains largely unchanged upon
adsorption.[41,42]Besides planarization,
the other significant consequence of adsorption
of ClB-SubPc molecules on Cu(111) is the change in their electronic
structure (Figure c). Notably, adsorption leads to a substantial “smearing”
of the molecule-projected DoS (PDoS) around the Fermi level of the
metal–molecule system. This occurs to the extent that different
molecular states are difficult to identify both in the PDoS and in
the measured ultraviolet photoelectron (UP) spectrum (see below).
The particularly strong smearing of the PDoS is caused by the hybridization
of occupied as well as unoccupied molecular states with the d-band
and the sp-band of copper. Similar observations have been made on
Cu substrates for other conjugated molecules, which are flat in the
gas phase. The relevant interfaces include, for example, pentacene
on Cu(111),[7,43] Cu(100),[44] and Cu(110),[45,46] as well as perylenetetracarboxylic
dianhydride (PTCDA) on Cu(111).[9,47]Additionally,
the ClB-SubPc monolayer is rendered metallic because
the Fermi level lies in a region of significant PDoS. Metallization
of the interface due to strong hybridization is a commonly observed
feature for planar molecules adsorbed on metal surfaces.[7,9,43−50] In fact, a previous investigation of ClB-SubPc on Cu(111) using
two-photon photoemission spectroscopy also suggested hybridization
and charge transfer to the molecular orbitals from metal surface states.[21]The strong broadening of the PDoS found
in the computational results
is also present in the experimental data. In Figure , we show a comparison of the UP spectra
of 0.7 monolayer equivalents (as defined in ref (22)) of ClB-SubPc on both
Cu(111) and HOPG. On HOPG, a distinct feature representing the ClB-SubPc
highest occupied molecular orbital (HOMO) centered at a binding energy
(EB) of −1.7 eV is clearly resolved
on top of the rather weak background originating from the HOPG π-band.
This is consistent with a UP spectroscopy report on a 7.5 nm thick
ClB-SubPc film on Si.[51] Further, peaks
are observed at higher binding energies. In contrast, on Cu(111) only
weak and broad features are visible between 0 and −2 eV. These
stem primarily from the Cu d-band He(I)β satellite
features at −0.8 eV. Additional contributions from the remainder
of the Cu(111) Shockley surface state (−0.2 to −0.4
eV) and from a partially filled lowest unoccupied molecular orbital
near the Fermi edge are expected at the positions indicated in Figure following the two-photon
photoemission data in ref (21). Below −2 eV, the Cu d-bands manifest as the strongest
features in the ClB-SubPc/Cu(111) UP spectrum. Their broad structure
is indicative of scattering in the organic adlayer and interfacial
hybridization. There is no clear evidence of the HOMO or other molecular
features in the UP spectrum, though a contribution from the HOMO to
the shoulder at EB = −2.2 eV or
the d-band region cannot be excluded (we do not necessarily expect
the HOMO feature on Cu(111) to be at the same energy as on HOPG).
In summary, the distinctly different contributions of ClB-SubPc to
the UP spectra when adsorbed on HOPG and Cu(111) support the notion
of a strong smearing of the molecular states at the ClB-SubPc/Cu(111)
interface that has been inferred from the simulations.
Figure 3
Ultraviolet photoelectron
spectra of a ClB-SubPc monolayer on Cu(111)
(top panel, black, strong hybridization) and HOPG (bottom panel, blue,
weak hybridization). The binding energy is referenced to the Fermi
level.
Ultraviolet photoelectron
spectra of a ClB-SubPc monolayer on Cu(111)
(top panel, black, strong hybridization) and HOPG (bottom panel, blue,
weak hybridization). The binding energy is referenced to the Fermi
level.The above observations raise the
question to what extent the particularly
strong coupling between substrate and adsorbate and the interaction-induced
planarization of most of the molecular backbone are interrelated.
To address this issue, as a first step we calculated the adsorption
energy Eads as a function of the adsorption
height hads. We define Eads aswhere Einter is
the energy of the interacting metal–molecule system, ECu is the energy of a pristine Cu(111) slab
(again, with only the top two layers relaxed), and Emol is the energy of a single relaxed ClB-SubPc molecule
in the gas phase. In addition, the methodology applied to treat van
der Waals interactions allows us to compute separately the contribution
of these energies to Eads, which we term
as EvdW (see SI for details).The adsorption height, hads, is defined
as the difference between the average heights of the C-18 atoms and
that of the metal atoms in the topmost layer of the reconstructed
substrate. At equilibrium, this height is 2.23 Å, which appears
low at first glance, but it is comparable to adsorption distances
measured for planar π-conjugated systems in similar situations,
such as the 2.34 Å found for pentacene on Cu(111).[52]In practice, to trace Eads vs hads, we gradually
moved the molecule from its
equilibrium position along the z-axis in steps of 0.2 Å, fixed
the position of the Cl atom, and then optimized the geometry of the
rest of the molecule as well as the top two atomic layers of the Cu
substrate. The resulting curve, shown in Figure a, displays a few peculiarities: (i) The
PBE contribution to the adsorption energy (green curve) is positive
for all distances. This hints at a distinctly repulsive interaction
between the ClB-SubPc molecules and the Cu(111) surface in the absence
of long-range van der Waals interactions. In fact, we observe a detachment
of the molecule from the surface, when van der Waals (vdW) interactions
are turned off and the system geometry is optimized. (ii) The dependence
of the adsorption energy on adsorption distance (when including vdW
attraction) displays an unusual, essentially linear shape between
the equilibrium distance of 2.23 Å and the largest considered
distance of 3.35 Å (r2 = 0.996 for
a linear fit plotted in orange). There is also no indication of multiple
minima, as are observed for systems displaying physisorbed as well
as chemisorbed structures.[6,53]
Figure 4
(a) Adsorption energy
(Eads) per ClB-SubPc
molecule as a function of adsorption distance including (black curve)
and neglecting (green curve) long-range van der Waals interactions
(EvdW). The orange line in the background
is a linear fit to the Eads (PBE + vdWsurf) curve, highlighting the unusual shape of the adsorption
potential. (b) Change in the planarity (Π) of the molecule as
a function of its adsorption height (red triangles). The pink dashed
curve indicates the “expected” trend extrapolated from
quadratic fitting of the data points with hads > 2.7 Å. The blue curve and the right axis refer to the
energetic
cost of planarizing the molecule (Πenergy), i.e.,
of changing the molecular geometry from that obtained for the gas
phase to that found at a specific adsorption distance.
(a) Adsorption energy
(Eads) per ClB-SubPc
molecule as a function of adsorption distance including (black curve)
and neglecting (green curve) long-range van der Waals interactions
(EvdW). The orange line in the background
is a linear fit to the Eads (PBE + vdWsurf) curve, highlighting the unusual shape of the adsorption
potential. (b) Change in the planarity (Π) of the molecule as
a function of its adsorption height (red triangles). The pink dashed
curve indicates the “expected” trend extrapolated from
quadratic fitting of the data points with hads > 2.7 Å. The blue curve and the right axis refer to the
energetic
cost of planarizing the molecule (Πenergy), i.e.,
of changing the molecular geometry from that obtained for the gas
phase to that found at a specific adsorption distance.Further insight can be gained by tracing the planarity
of the adsorbed
molecules as a function of hads, as shown
in Figure b (red curve):
First, Π increases smoothly, as the molecule approaches the
surface. However, at ca. 0.2 Å above the equilibrium distance,
an increased planarization sets in, as is particularly evident when
comparing the actual behavior at short distances with an extrapolation
from larger adsorption heights (i.e., for hads ∈ [2.79, 3.35 Å], pink dashed curve). This means that
the molecule undergoes a disproportionately large planarization within
a very small range of distances (∼0.2 Å) above the equilibrium
height. This happens in spite of the significantly increased energetic
cost to do so, as is evidenced by the calculated planarization energy
in Figure b (blue
curve).
Uncovering the Interplay between Adsorption
and Planarization
Given the unusual behavior of the ClB-SubPc/Cu(111)
interface (massive molecular distortion, strong hybridization, and
unusual shape of the adsorption potential), we focus next on the role
of the vdW forces. Here, we started from the repulsive situation,
i.e., using solely the PBE functional (in the absence of long-range
vdW interactions). Then, we introduced and gradually increased the
long-range van der Waals attraction, in this way pulling the molecule
toward the surface. This scaling of the van der Waals forces allowed
us to follow the gradual transition from an essentially noninteracting
metal–molecule system to a situation characterized by a significant
distortion of the adsorbate geometry and a strong hybridization between
the molecular and metal states.Technically, the rescaling is
achieved by calculating the energy contribution due to long-range
van der Waals interactions, analogous to the original Tkatchenko–Scheffler
scheme,[54] while introducing a scaling prefactor S that varies between 0 and 1. S scales
the C6,AB coefficients that govern the
pairwise interatomic interactions between two atoms A and B. This
yieldsHere, RAB is the
interatomic distance and fAB is a Fermi-type
damping function that subdues the correction at short range. Notably,
the Tkatchenko–Scheffler (TS) approach, like Grimme’s
D3 scheme,[55] the Becke–Johnson scheme,[56] and the scheme from Sato and Nakai,[57] aims to include the influence of the atom-pair’s
local chemical environment into the C6 coefficients. In TS, this is achieved by making the C6 coefficient dependent on the charge density around the
atoms to account for the dependence of vdW interactions on the system’s
local polarizability.[58] This makes the
TS approach particularly useful in the present case, as in the course
of changing S the charge distribution at the interface
undergoes pronounced changes (vide infra).The resulting van
der Waals contribution to the adsorption energy EvdW as a function of S is shown
in Figure a (black
curve). Notably, the curve does not increase smoothly with S, in contrast to the evolution of the van der Waals energy
for an isolated molecule (green curve). Rather, there are more pronounced
drops between 0.2 ≤ S ≤ 0.4 and between
0.6 ≤ S ≤ 0.8. These more abrupt changes
in energy are accompanied by stepwise increases in the planarity of
the ClB-SubPc molecules (Figure b) and by stepwise decreases in the adsorption heights
(Figure c). Consequently,
when scaling the vdW interactions and in this way pulling the molecule
toward the surface, one observes two clearly distinct adsorption regimes:
Between S = 0.4 and 0.6, the molecule adsorbs in
a somewhat planarized geometry (Π ≈ 49%) at a comparably
large distance of around 3.0 Å. Conversely, for S ≥ 0.8, the C-18 atoms of ClB-SubPc become essentially planar
(Π ≈ 94%) and the adsorption height reduces to 2.2 Å.
The transition between these regimes coincides with massive modifications
of the electronic structure of the interface, as discussed in the
following section.
Figure 5
Evolution of the properties of the ClB-SubPc/Cu(111) interface
with increasing the C6 scaling prefactor S. (a) Nonlinear increase in EvdW (the contribution of the long-range van der Waals interaction to
the adsorption energy) as a function of the van der Waals scaling
prefactor S (black curve) and linear increase in EvdW vs S for an isolated molecule
in vacuum (green curve). (b) Molecular distortion quantified by the
planarity of the adsorbed molecule (Π), as defined above. (c)
Calculated adsorption height, hads. There
are two distinct regimes with essentially constant planarity and adsorption
height (S = 0.4 to 0.6 and 0.8 to 1.0), shaded
in yellow and orange, respectively.
Evolution of the properties of the ClB-SubPc/Cu(111) interface
with increasing the C6 scaling prefactor S. (a) Nonlinear increase in EvdW (the contribution of the long-range van der Waals interaction to
the adsorption energy) as a function of the van der Waals scaling
prefactor S (black curve) and linear increase in EvdW vs S for an isolated molecule
in vacuum (green curve). (b) Molecular distortion quantified by the
planarity of the adsorbed molecule (Π), as defined above. (c)
Calculated adsorption height, hads. There
are two distinct regimes with essentially constant planarity and adsorption
height (S = 0.4 to 0.6 and 0.8 to 1.0), shaded
in yellow and orange, respectively.
Interfacial Electronic Structure
A first impression of modifications to the electronic structure of
the ClB-SubPc/Cu(111) interface caused by increasing the van der Waals
attraction can be obtained from the adsorption-induced changes in
the work function, ΔΦ. For the sake of analysis, this
quantity can be separated into a contribution from the monolayer due
to molecular dipoles, ΔΦML, and a contribution
from the so-called “bond-dipole” resulting from the
interaction between substrate and adsorbate, dominated by the adsorption-induced
charge rearrangements, ΔΦBD.[59] This yieldsIn practice, ΔΦ
and ΔΦML are calculated explicitly, and ΔΦBD is obtained as the difference between these two numbers.
Note that
the above quantities for the subsystems are calculated in the optimized
geometries they adopt upon adsorption. The change in work function
for a given system is obtained as the energetic difference between
the vacuum levels above and below the calculated slab (for more details
see ref (59)).As shown in Figure , for small S, ΔΦ is positive. This
is a consequence of the significant molecular dipole moment pointing
toward the surface, which results in a large positive value of ΔΦML. As the adsorption distance for small S is large, the change in work function due to the interaction-induced
charge rearrangements, ΔΦBD, has only a minor
impact. This changes in the first adsorption regime (for 0.4 ≤ S ≤ 0.6), where the absolute value of ΔΦBD rises by 0.5 eV. Simultaneously, the increasing planarity
of the adsorbed molecules results in a decreasing molecular dipole
and, thus, a smaller ΔΦML. (The values for
the molecular dipoles of the different geometries obtained when varying S can be found in the SI.) As
both trends act in the same direction, the overall effect is a change
in the sign of the work function modification and a resulting ΔΦ
of −0.2 eV. This value remains constant over the first adsorption
regime.
Figure 6
Evolution of the change in work function, ΔΦ (black),
the contributions due to the molecular layer, ΔΦML (red), and the interfacial charge rearrangements, ΔΦBD (blue), with increasing C6 scaling
prefactor S. The work function of the pristine Cu(111)
surface is calculated to be 4.8 eV.
Evolution of the change in work function, ΔΦ (black),
the contributions due to the molecular layer, ΔΦML (red), and the interfacial charge rearrangements, ΔΦBD (blue), with increasing C6 scaling
prefactor S. The work function of the pristine Cu(111)
surface is calculated to be 4.8 eV.Upon approaching the second adsorption regime, another step
in
the work function modification is observed and ΔΦ drops
to ∼−0.5 eV (for 0.8 ≤ S ≤
1.0). This is primarily a consequence of the additional planarization
of the adsorbate molecules resulting in a further reduced molecular
dipole (see SI), such that ΔΦML becomes ∼0.06 eV. Notably, the contribution to the
work function modification arising from the molecule–metal
interaction changes much less, and ΔΦBD saturates
at −0.55 eV. This is quite surprising, considering the massive
change in the adsorption distance and the fundamentally different
shapes of the interfacial charge rearrangements between the first
and second adsorption regimes (vide infra).A direct comparison
of the calculated ΔΦ to experiments
is difficult, as the experimentally accessible interface contains
multiple species,[22] including dechlorinated
molecules (for which we calculate a ΔΦ of −0.8
eV). Nevertheless, it is interesting to point out that the experimentally
determined work function modification at a nominal coverage of 0.5
monolayer equivalents (corresponding to the coverage in the simulations)
amounts to −0.5 eV.[21]The
evolution of ΔΦBD can be understood
on the basis of the changes in the energetic alignment between the
electronic states in the metal, characterized by the position of the
Fermi level, and the states in the adsorbate layer. This is shown
in Figure .
Figure 7
Density of
states projected onto the molecular region (PDoS) for
the ClB-SubPc/Cu(111) interface as a function of the van der Waals
scaling prefactor S. The DoS of the gas-phase molecule
has been Gaussian broadened and shifted such that the vacuum level
of the molecule is the same as that of a clean Cu slab. The Fermi
level of that slab is then used as the energy reference for the isolated
molecule in the above plot. This corresponds to assuming vacuum-level
alignment and describes the situation for an isolated molecule far
away from the Cu surface, i.e., in the absence of electronic interactions
between substrate and adsorbate. The shading denotes the different
adsorption regimes (cf., Figures and 6).
Density of
states projected onto the molecular region (PDoS) for
the ClB-SubPc/Cu(111) interface as a function of the van der Waals
scaling prefactor S. The DoS of the gas-phase molecule
has been Gaussian broadened and shifted such that the vacuum level
of the molecule is the same as that of a clean Cu slab. The Fermi
level of that slab is then used as the energy reference for the isolated
molecule in the above plot. This corresponds to assuming vacuum-level
alignment and describes the situation for an isolated molecule far
away from the Cu surface, i.e., in the absence of electronic interactions
between substrate and adsorbate. The shading denotes the different
adsorption regimes (cf., Figures and 6).The trend for ΔΦBD from Figure is expected to map directly
onto the evolution of the level alignment, as demonstrated for a variety
of organic molecules on Ag(111).[60] Indeed,
up to S = 0.4, one sees a pronounced shift in the
states of the ClB-SubPc monolayer toward higher binding energies (by
ca. 0.7 eV). This can be explained by Pauli pushback[61] reducing the surface dipole of the metal[1] and, thus, its work function. Consistent with a constant
adsorption height and adsorbate shape, there is no further change
in the PDoS up to S = 0.6. Notably, there is virtually
no difference in the shape of the PDoS compared to the effectively
noninteracting situation (S = 0). Upon further increasing S (i.e., upon pulling the molecule further toward the surface),
one again observes a shift of the PDoS toward higher binding energies
by 0.4 eV. However, like in the above-discussed evolution for ΔΦBD, the net effect is weaker than what is expected based on
the strongly decreasing adsorption distance and the resulting massively
amplified Pauli pushback. The reason for that is that here one enters
the regime of Fermi-level pinning: The formerly unoccupied PDoS of
the adsorbate layer is shifted into resonance with the Fermi level,
resulting in electron transfer from the metal to the molecules such
that the resulting dipole counteracts the effect of Pauli pushback
and the overall interfacial dipole remains largely constant.[62]On the basis of these observations, one
can denote the first adsorption
regime (0.4 ≤ S ≤ 0.6) as the “weak-coupling
regime” with only negligible charge transfer between molecular
states and the metal, as is usually observed for physisorption. Here,
it should be stressed that this regime is not merely a curiosity resulting
from the scaling of the van der Waals attraction. In fact, it is representative
of the situation of ClB-SubPc adsorbing on less reactive metals, as
is shown in the SI for the ClB-SubPc/Ag(111)
interface, or on other even more inert substrates like HOPG (see Figure ).The second
regime (0.8 ≤ S ≤ 1.0)
is then denoted as the “strong-coupling regime”, as
there we see the partial filling of the molecular PDoS suggesting
that electrons have been transferred from the Cu substrate into previously
unoccupied orbitals of the molecular layer. Moreover, a massive broadening
of all molecule-derived features in the PDOS occurs. As mentioned
earlier, the latter is indicative of a particularly strong hybridization
between molecular and metal states. However, we refrain from associating
it with chemisorption, as in spite of the significant interaction
between the ClB-SubPc molecules and the Cu(111) substrate, Figure a suggests that there
would be no bonding in the absence of van der Waals attraction. That
is, we face a situation of strong electronic coupling between substrate
and adsorbate that by itself would, however, not be sufficient to
trigger adsorption.To further distinguish between the two adsorption
regimes and to
trace the strong hybridization between substrate and adsorbate states,
it is useful to consider the adsorption-induced interfacial charge
rearrangements, Δρ. These are obtained viawhere ρinter is the charge
density of the interacting metal–molecule system, ρslab is the charge density of only the metal slab, and ρmol is the charge density of a molecule in the isolated monolayer
(again, all in the geometry obtained for the interacting system).
Δρ can be further averaged/integrated along one or two
spatial directions to obtain two-dimensional (2D) or one-dimensional
(1D) charge rearrangement plots, respectively.The situation
for the weak-coupling regime (S =
0.4) is shown in Figure a. The charge rearrangements indicate the dominance of Pauli pushback
with a pronounced electron-density depletion in the region of the
overlap between the tailing electron cloud of the metal and the π-states
of the molecules. The effect is stronger below the plane of the molecule,
as there the tailing states from the metal are more prevalent than
above. The “pushed-back” electrons are accumulated directly
above the metal surface (see also plot of plane-integrated 1D charge
rearrangements contained in the SI).
Figure 8
Side view of
the interfacial charge rearrangements for (a) S =
0.4 and (b) S = 1.0 averaged over the
unit cell along the direction perpendicular to the plane of projection.
Side view of
the interfacial charge rearrangements for (a) S =
0.4 and (b) S = 1.0 averaged over the
unit cell along the direction perpendicular to the plane of projection.The situation changes qualitatively
in the strong-coupling regime,
shown for S = 1.0 in Figure b. There, in addition to the electron-density
depletion immediately below the molecular backbone, one observes accumulation
directly above the molecule and, even more pronounced, in the region
about halfway between the C-18 atoms and the topmost layer of
the Cu(111) substrate. Especially, the latter deserves attention:
As shown in Figure a, electron accumulation occurs in those regions where the C-18 atoms
and the Cu atoms of the topmost metal layer are particularly close,
whereas there is depletion above the bridge and hollow sites of the
substrate. The assessment that the driving force for electron accumulation
in specific regions is the local proximity between the molecular π-system
and on-top metal sites is confirmed by the very different shape of
the charge rearrangement pattern for ClB-SubPc on the fourfold symmetric
Cu(100) surface shown in Figure b.
Figure 9
Top view of the charge rearrangements induced by ClB-SubPc
on (a)
Cu(111) and (b) Cu(100). The graphs show a 2D slice of the charge
rearrangements calculated for the plane halfway between the molecule
and the metal.
Top view of the charge rearrangements induced by ClB-SubPc
on (a)
Cu(111) and (b) Cu(100). The graphs show a 2D slice of the charge
rearrangements calculated for the plane halfway between the molecule
and the metal.We also find that the
H atoms are located 0.2 Å above the
C-18 plane and that the carbon–carbon bond lengths within the
rings constituting the C-18 system increase on average by ∼3
pm. These are further indications for a minor disruption of the molecular
π-system caused by the hybridization of the molecular and metal
states.
Role of Molecular Planarity in its Ability
to Hybridize
Even though the above results provide evidence
that the planarization of the ClB-SubPc molecule is an integral aspect
of the adsorption process, the data presented thus far do not fully
clarify how crucial planarization actually is for the electronic structure
of the interface and especially for the strong hybridization of molecular
and metal states. To address this question, we simulated the adsorption
of a rigidified ClB-SubPc molecule. For this, we fixed the lateral
positions of the carbon atoms in the gas-phase geometry of the molecule,
thereby reducing the degrees of freedom during geometry relaxation.
In other words, the C atoms could relax solely along the direction
perpendicular to the surface, thereby largely preventing a planarization
of the molecule. Care was taken to align the rigidified molecule such
that its registry with the Cu(111) surface was identical to that of
the fully relaxed case. The optimum adsorption height for the rigidified
molecule rose to 3.0 Å, 0.77 Å more than the fully relaxed
structure (vide supra). Interestingly, this adsorption height is almost
identical to that obtained for the fully optimized interface in the
weak-coupling regime (calculated in the full optimizations for 0.4
≤ S ≤ 0.6). Therefore, in Figure a, we compare the
PDoS of the fully optimized and rigidified structures (both calculated
at S = 1.0) with the situation obtained for the weak-coupling
case (S = 0.4, fully optimized).
Figure 10
(a) Comparison of the
molecule-projected densities of states for
the artificially rigidified molecule (S = 1.0) and
the fully relaxed structures in the weak (S = 0.4)-
and strong (S = 1.0)-coupling regimes. (b) Side view
of the interfacial charge rearrangements for the rigidified molecule
averaged over the unit cell along the direction perpendicular to the
plane of projection.
(a) Comparison of the
molecule-projected densities of states for
the artificially rigidified molecule (S = 1.0) and
the fully relaxed structures in the weak (S = 0.4)-
and strong (S = 1.0)-coupling regimes. (b) Side view
of the interfacial charge rearrangements for the rigidified molecule
averaged over the unit cell along the direction perpendicular to the
plane of projection.Intriguingly, we find an almost perfect correspondence between
the latter two cases in the PDoSes as well as in the work function
modifications (ΔΦ = −0.14 eV for the rigidified
molecule and −0.17 eV for weak-coupling regime). Likewise,
the charge rearrangements for the rigidified system show no indication
of strong hybridization (Figure b). Rather, one solely observes Pauli pushback, as
displayed already in Figure a for the weak-coupling regime. This clearly shows that the
ability of the nonplanar ClB-SubPc to strongly couple to the metal
substrate crucially depends on its ability to planarize, which occurs
only when van der Waals attractions pull the molecule toward the surface.
Summary and Conclusions
The ClB-SubPc/Cu(111)
interface provides valuable insight into
how changes in an adsorbate geometry induced by the van der Waals
attraction to a substrate can change the electronic coupling between
substrate and adsorbate, pushing the interface into a significantly
different coupling regime. To better understand the peculiarities
of this interface, we adopted a strategy that allowed us to scale
the van der Waals interactions, to gradually pull the molecule toward
the substrate. This revealed two adsorption regimes, which we identified
to be cases of weak and strong interfacial coupling. In the former,
the originally shuttlecock-shaped molecules are planarized to a lesser
degree, and the electronic interaction with the substrate is dominated
by Pauli pushback. This, together with the molecular dipoles, determines
the adsorbate-induced work function modification. In the second regime,
the molecule is pulled very close to the surface, it is largely planarized,
and a particularly strong hybridization between molecular and metal
states occurs in conjunction with electron transfer from the substrate
to adsorbate. In this case, the adsorbate layer becomes metallic,
and the work function is determined by Fermi-level pinning. Notably,
we show that this regime cannot be characterized as conventional chemisorption,
as in spite of the sizable electronic interactions between substrate
and adsorbate, the molecule would not attach to the surface in the
absence of van der Waals attraction. In fact, the strong-coupling
regime can be reached only provided that the molecule significantly
changes its shape, which is shown to be a consequence of van der Waals
forces pulling it toward the surface. Thus, the present system can
be regarded as an intriguing example of a van der Waals-enabled strong
electronic coupling between a metal surface and an organic adsorbate
molecule.
Authors: Nahid Ilyas; Shashank S Harivyasi; Percy Zahl; Rocio Cortes; Oliver T Hofmann; Peter Sutter; Egbert Zojer; Oliver L A Monti Journal: J Phys Chem C Nanomater Interfaces Date: 2016-03-10 Impact factor: 4.126
Authors: Elisabeth Wruss; Oliver T Hofmann; David A Egger; Elisabeth Verwüster; Alexander Gerlach; Frank Schreiber; Egbert Zojer Journal: J Phys Chem C Nanomater Interfaces Date: 2016-02-29 Impact factor: 4.126