Postadsorption Work Function Tuning via Hydrogen Pressure Control

The work function of metal substrates can be easily tuned, for instance, by adsorbing layers of molecular electron donors and acceptors. In this work, we discuss the possibility of changing the donor/acceptor mixing ratio reversibly after adsorption by choosing a donor/acceptor pair that is coupled via a redox reaction and that is in equilibrium with a surrounding gas phase. We discuss such a situation for the example of tetrafluoro-1,4-benzenediol (TFBD)/tetrafluoro-1,4-benzoquinone (TFBQ), adsorbed on Cu(111) and Ag(111) surfaces. We use density functional theory and ab initio thermodynamics to show that arbitrary TFBD/TFBQ mixing ratios can be set using hydrogen pressures attainable in low to ultrahigh vacuum. Adjusting the mixing ratio allows modifying the work function over a range of about 1 eV. Finally, we contrast single-species submonolayers with mixed layers to discuss why the resulting inhomogeneities in the electrostatic energy above the surface have different impacts on the interfacial level alignment and the work function.


A. Supercell Geometries
In this chapter we describe the particular supercell geometries that were investigated on Cu(111) and Ag(111). Due to the single bond of the OH-group of the TFBD molecule to its phenyl ring it is freely rotatable along the bonding direction. By rearranging the TFBQ and TFBD molecules on the adsorption sites with gradually increasing TFBD fraction, f, and different alignments of the rotatable OH-groups, we found that for the ground state energy the alignment of the OH-bonds is crucial. To converge into the global energy minimum in the geometry optimization procedures it is thus necessary to rotate those OH-groups in such a way that the hydrogen comes in the vicinity of an oxygen atom of a neighboring TFBQ molecule. This then leads to the formation of a hydrogen bond. It turns out that four geometries, as sketched in Figure S1 serve as basic building blocks for the supercell geometries. To identify a certain supercell geometry based on its building blocks we introduced the following nomenclature. Starting from the lower, left adsorption site of a supercell the respective sub-geometries shown in Figure S1 are labelled by the numbers 1, 2, 3 and 4. A new row is indicated by a slash symbol "/". The supercell geometry shown in Figure S1 of the main paper then for instance would be labelled as "21/12". By proper alignment of those building blocks a set of supercell geometries were created that served as starting points for geometry optimizations. A list of the supercell geometries that were investigated within this work is quoted in Table S1. The corresponding relaxed supercell geometries of mixed monolayers and sub-monolayers are shown for adsorption upon Cu (111) in Figure S2 and for Ag(111) in Figure S3.
Depending on the neighboring molecules above and below a unit cell for TFBD in the adsorption site two alignments of the OH-bonds minimize the ground state energy: Geometry 2 is preferable if hydrogen bonds are formed to neighboring TFBQ molecules --the OHgroup is rotated such that the hydrogen approaches the oxygen atom of the neighboring TFBQ molecule as close as possible. For neighboring TFBD molecules geometry 3 minimizes the ground state energy -on Cu(111) for instance E 33/33 -E 22/22 = -0.63 eV. (See also discussion in the main text) In mixed layers of electron-donating and -accepting molecules or SAMs cases of superstructure formation and phase separation were reported. 1,2 To check if phase separation is likely to occur for the investigated system, a set of supercell geometries with different arrangements of TFBQ and TFBD molecules at 50 % TFBD fraction were evaluated. We found that checkerboard motifs are always more stable than structures where molecules are arranged in rows or clusters. However the difference in the total energies is just a few meV: on Cu(111) for instance E 12/21 -E 12/12 = -4.5 meV and E 12/21 -E 11/22 = -1.5 meV. This shows that alternating patterns are energetically favorable, and hence, for a mixed monolayer of TFBQ and TFBD phase separation is not expected.

B. Sub-Monolayers of TFBD on Ag(111)
The contributions of the molecular and bond dipole to ∆Φ for sub-monolayers of TFBD on Ag(111) are displayed in Figure S5. Analyzing the evolutions one finds that the deviations from linearity mainly stem from the evolution of the molecular dipole, whereas the contributions from the bond dipole nearly follows a linear evolution (in fact, slightly parabolic coverage dependency) of the bond dipole.

C. Adsorption Heights and Charges in Mixed Monolayers
For Cu(111) and Ag(111) the adsorption heights are shown in S6a and Figure S6b.
Comparing the adsorption heights for TFBD in mixed monolayers (blue filled diamonds) to sub-monolayers (blue diamonds) one finds that for both coinage metals TFBD comes closer to the surface the larger the number of neighboring TFBQ molecules is. The decrease of the adsorption height in mixed monolayers likely stems from the electrostatic attraction with TFBQ. Conversely the TFBQ molecules increase their adsorption heights only on Cu(111).
On Ag (111), where the adsorption height of a homogeneous TFBQ monolayer is about 0.4 Å larger than on Cu(111), the mean adsorption height stays approximately constant independent of f, which is also the case for the corresponding sub-monolayer case. Such an "equalization of the adsorption heights" was also observed experimentally for different mixed-monolayers of electron-accepting and -donating species. 3 A further effect that was mentioned there was the modification of the electron spill out due to the adsorption of the molecules that affects neighboring adsorption sites. However, for the system investigated here, this is a minor effect.  We remark that in hypothetical, free-standing mixed monolayers charge-transfer from TFBD to TFBQ occurs, and that the TFBD-HOMO is found to be in resonance with the TFBQ-LUMO. Since there is no appreciable overlap of the wave functions in the coplanar monolayer, we speculate that this charge-transfer might be a spurious artefact of the PBE functional, akin to the situation discussed for other donor/acceptor pairs. 5,6 Interestingly, the HOMO/LUMO resonance situation is lifted when the mixed layers are adsorbed onto the surface. This yields a situation that is in better accordance with chemical intuition and supports our earlier claim that PBE+vdW surf yields reasonable electronic structures for metal/molecule interfaces. 7

D. A different viewpoint on Fermi-level pinning in mixed layers
The potential caused by the TFBD molecules explains the differences of the ∆Φ(f)-evolutions on Cu and Ag. Due to the smaller adsorption height of TFBQ on Cu the distance to neighboring TFBD molecules is larger -compare situation 3 sketched in Figure S8a. The net charge formed at the TFBD-adsorption site due to the Pauli-pushback effect with the mirror charges in the metal lead to an electric field that weakly affects the energy of the TFBQ-LUMO due to its spatial position. For Cu therefore a smaller downshift of the LUMO (blue arrow pointing down in Figure S8b) and thus a smaller compensation from the bond dipole occurs (red arrow pointing up in Figure S8b). For mixed monolayers adsorbed upon Ag on the other hand the adsorption height of TFBQ is by about 0.45 Å larger than on Cu -compare situation 2 sketched in Figure S8a. Therefore the electric field shifts the TFBQ-LUMO strongly down in energy (blue arrow pointing down in Figure S8b) what results in a strong compensation by the bond dipole (red arrow pointing up in Figure S8b). This further is reflected in the weaker decrease of ∆Φ(f).
Note that within this idealized picture the LUMO-peak is modelled as a Delta-Distribution.
However, due to the interaction of an adsorbed molecule with the adsorbent and neighboring molecules the molecular energy levels broaden. This causes that the Fermi-level-pinned LUMO-peak is not directly located at E F − as sketched in Figure S8.

E. Infra-red-spectra of TFBQ and TFBD
To assess the quality of our calculations for the vibrational properties of the molecules, we calculated the the Infra-red (IR) spectra of TFBQ and TFBD in the gas phase (shown in figure   S9). For comparison also the corresponding IR-spectra from the NIST database 8 are given.
The IR spectrum calculated via DFT shows overall good agreement with reference data from experiment. However, vibrations at intermediate wavenumbers (<1200 cm −1 ) are calculated at slightly lower energies than the peaks in the reference data set, while for very high wavenumbers (>3000 cm −1 ), the calculated energies are slightly too large. Interestingly, for the TFBD IR-spectrum a peak at 377 cm −1 is predicted by our calculations that does not appear in the reference data set.

F. Extraction of Metal-Atoms due to TFBQ adsorption
The adsorption of organic molecules on metal surfaces often perturbs the electronic structure of the surface atoms, in particular if a (partially covalent) bond is formed. As a result, the surface metal atoms often rearrange. The extent of these rearrangements varies from relatively small (e.g., ca. 0.3 Å for the adsorption of TCNQ on Cu 9 ) to significantly more than 1.0 Å (TCNE on Cu 10 ). In several cases, even the formation of surface adatoms has been observed. 11,12 For large molecules bearing carboxyl groups (i.e., PTCDA), however, the effects is expected to be generally small. 13 In this work, we observe larger surface atom rearrangements for the bond-forming TFBQ molecule than for the inert TFBQ. However, even there the rearrangements are relatively unspectacular. The difference in the mean zposition of the extracted metal atoms to the z-position of the other atoms in the uppermost metal layer is 0.13 Å on Cu(111) and 0.05 Å on Ag(111).