Electronic properties: work function modification, charge differ-

Proceeding with the discussion of the electronic properties of the dd 00 00 structure, we first analyze the change of the work function △Φ brought about by the adsorption of the oxalic acid molecules onto the clean Cu(011) surface. For the clean Cu(011) surface we had a work function of Φ = 4.421 eV (see Chapter 3), for the system obtained after the adsorption of the oxalic acid in the dd 00 00 conformation we find a work function of about Φ = 7.822 eV. This increase of the work function can be easily seen in Figure 4.7 where we have plotted the local potential as a function of positions perpendicular to the surface (x -direction). In order to demonstrate the different work functions of the clean surface and of the surface with adsorbed oxalic acid we have extended the asymmetric slab one more time in the x -direction. Note that on the left hand side of diagram 4.7a the clean Cu(011) surface is shown (unreconstructed bulk terminated) with a work function Φ1 = 4.368 eV, and on the right hand side the adsorption layer with a work function Φ2

= 7.822 eV. When the reconstruction of the clean surface is taken into account, the work function changes to Φ1 = 4.421 eV. Note also that the local potential is not continuous at the boundary of the supercell. Due to the asymmetry of the cell with two different

4.1 The dd 00 00 structure 91 surfaces a dipole exists at the boundary, which has been taken into account when calcu-lating the total energy of supercells in the VASP code.

The modification of the work function, after the adsorption process, is caused by the charge reorganization at the molecule-surface interface. More precisely, it gives us infor-mations about the direction and extent of the surface-adsorbate electronic charge transfer.

Typically, a Φ increase implies substrate to adsorbate charge transfer, while a Φ decrease implies adsorbate to substrate charge transfer. Applying this rule of thumb to our system, we deduce that electrons are transferred from the Cu surface atoms to the O atoms, due to the adsorption. This is definitely very reasonable given that the electronegativity of Cu is of only 1.9 while the one of O is 3.5. As a consequence of this charge transfer we get an excess of negative charges on the outside and an excess of positive charges on the inside of the surface. This leads to a positive dipole (pointing outward) that reinforces the original surface dipole (due to the common phenomenon of electron “spill out“ at any surface) causing thus an increase of the work function.

The magnitude of this dipole can be found using the Helmholtz equation which reads:

µ_x = ǫ0

e · A · ∆Φ (4.2)

where ǫ0 = 5.52635 · 10⁻³· ^e

V ·˚_A is the permittivity of the vacuum, e the elementary charge, A the surface area (in ˚A²), ∆Φ the work function difference (in eV) and µ_xthe component of the dipole moment directed along the surface normal (in electron-Angstr¨om e˚A). (An alternative formula is µx = (_12·π¹ ) · A · ∆Φ where the symbols have the same meaning as before but the dipole moment is obtained in Debye (D) instead of e˚A). According to this equation the dipole moment along the surface normal for the dd 00 00 structure has a magnitude of 0.702 e˚A(3.376 D). The dipole moment obtained this way can be further expressed as the sum of the x-component of the dipole moment of the molecular monolayer (µM L) in the absence of the copper substrate and the x-component of the dipole moment created through the charge transfer/reorganization that accompanies the chemical bond formation between the molecules and the surface (µbond), according to:

µx = µx,bond+ µx,M L (4.3)

In order to calculate µ_{x,M L} we have used the ML geometry as obtained from the op-timized structure of the ML adsorbed on the Cu(011) surface. We point out that this means that we have used the radical COO instead of the less reactive COOH-COOH molecule. This was done in order to avoid the loss of generality of the analysis which would follow in the case a certain position for the H atom would have been chosen.

With this approach we obtain a value of -0.210 e˚A(-1.009 D) for the µx,M L. This means that the ML-dipole is directed from the vacuum to the surface (pointing inward). On the contrary, the total dipole µx= 0.702 e˚A is pointing outward. We conclude that the largest contribution to the work function increase comes from the interface dipole and not from the inner ML dipole.

A visual impression of the interface dipole can be gained by plotting the layer resolved charge density difference. We define the charge density difference ρ_{dif f} through:

ρdif f = ρ − (ρ^Cu(011)− ρ^{M L}) (4.4)

Figure4.7:(a):determinationoftheworkfunctionbycomparingthevacuumpotentialwiththeFermilevel.NoticethedifferencebetweentheworkfunctionofthecleanCu(011)surface(Φ1)andthatoftheMLcoveredCu(011)surface(Φ2).(b):lateralviewofthelocal(electrostatic)potentialinthe(1

1 ¯1)plane.(c):lateralviewofthelocalpotentialinthe(0

ofthelocalpotentialinthe(0 1 ¯1)plane.(d):almosttopview

molecules. 1 ¯1)plane.Inallthecasesthedeepminimaarelocatedatthetwocarboxylicgroupsoftheoxalicacid

4.1 The dd 00 00 structure 93 where ρCu(011) and ρM L denotes the charge density of the clean Cu(011) surface and of the isolated oxalic molecular layer, respectively, both with the geometry as obtained from the optimized structure of the ML adsorbed onto the Cu(011) surface. By averaging the charge density in planes parallel to the Cu(011) surface, we obtain a one-dimensional rep-resentation of the charge density. From the inspection of this plot (Fig. 4.8) we find that a series of alternating regions of charge accumulation (ρ_{dif f} > 0) and charge depletion (ρdif f < 0) (i.e. dipoles) appears due to the adsorption. As expected, the strongest dipole occurs along the Cu-O bond and it is due to a charge transfer of about 0.4e⁻ from the Cu surface to the ML. This dipole is the one which appears to be mainly responsible for the work function increase. A compensating contribution comes definitely also from the opposite variation of the charge inside the ML.

An interesting consequence of the substrate-ML charge transfer is the decrease of the hole injection barrier from the Cu(011) towards the ML. This follows from the fact that a Cu-ML charge transfer results in a lowering of the Fermi level (or an increase of the work function) on the metal side and a slight increase of the HOMO energy on the ML side.

If the charge transfer would have taken place from the ML towards the substrate then it would have been the electron injection barrier -or the Schottky barrier- which would have been reduced. Both cases are schematically illustrated in Fig. 4.9 (upper panel).

These aspects appear clearly when comparing the total DOS plots for the clean Cu(011) surface, ML in vacuum and the ML adsorbed onto the Cu(011) surface (see Fig. 4.9).

We see then that the hole injection barrier is, after adsorption, about 0.664 eV while the Schottky barrier (i.e. the electron injection barrier) is about 2.966 eV. It is clear from these numbers that the electron current is almost completely suppressed following the adsorption, while the hole injection is sensibly favored. Such a behavior might be useful in designing certain optoelectronic devices (e.g. photovoltaic devices) for which it is important to fully block one type of charge carriers (in this case the electrons).

The investigation of the gap region between HOMO and LUMO (Fig. 4.10) reveals a clearly non-zero molecular LDOS in this energy region. Thus, similar to the known metal-semiconductor junction case, the interaction between the metal and the ML gives rise to what one might call metal-induced-gap-states (MIGS). The intensity of the MIGS peaks is smaller by about one order of magnitude than the average intensity of the ML PLDOS. However, this is somehow to be expected as these are states in the gap domain.

For smaller gaps these states can play an important role in electron tunneling between the HOMO and the LUMO states.

Figure4.8:(a):Electrostaticpotentialandchargedensitydifference;(b):charge(notdensity)differenceplot.Positionofrelevantatomsisgiventhroughdashedlinesandatomicsymbols.Noticethesizeofthechargetransferattheinterface;(c):3Dchargedensitydifferenceplot.Blue-depletion,red-accumulation;(d):theLaplacianofthechargedensitydifference.Noticetheorbitals(ofCuorO)involvedindonation/receptionofcharge.

4.1 The dd 00 00 structure 95

Figure 4.9: Up: sketch of the impact of the increased/decreased of the interface dipole on the electronic levels at the metal/organic interface. When the interface dipole is increased the hole injection barrier (HIB) is decreased and the electron injection barrier (EIB) increased. When the interface dipole is decreased the opposite effect occurs. Down:

comparative plots of LDOS for the clean Cu(011), ML in vacuum and ML adsorbed onto Cu(011). Fermi level - red lines, HOMO/LUMO - blue lines. The zero value is put at the vacuum potential energy of the 3 systems. The vacuum value is taken to be the one from the right(relaxed Cu surface)/left(at the deprotonated group of the ML)/right(molecular layer on Cu) side of the systems).

-4 -3 -2 -1 0 1 2 3 4

Figure 4.10: Up-left: LDOS for ML and Cu atoms; Up-right: same as before but em-phasizing now the region in the immediate vicinity of the Fermi level. The electron/hole injection barrier are also schematically displayed. In the inset a clearer image of the molecular MIGS is given. Middle-left: Plot of the spatial variation of HOMO for the ML in vacuum (with the same geometry as adsorbed), Middle-right: Plot of the spatial varia-tion of HOMO for the adsorbed ML. Down-left: Plot of the spatial variavaria-tion of LUMO for the ML in vacuum, Down-right: Plot of the spatial variation of LUMO for the adsorbed ML.