Employing X-ray Photoelectron Spectroscopy for Determining Layer Homogeneity in Mixed Polar Self-Assembled Monolayers.

Self-assembled monolayers (SAMs) containing embedded dipolar groups offer the particular advantage of changing the electronic properties of a surface without affecting the SAM-ambient interface. Here we show that such systems can also be used for continuously tuning metal work functions by growing mixed monolayers consisting of molecules with different orientations of the embedded dipolar groups. To avoid injection hot-spots when using the SAM-modified electrodes in devices, a homogeneous mixing of the two components is crucial. We show that a combination of high-resolution X-ray photoelectron spectroscopy with state-of-the-art simulations is an ideal tool for probing the electrostatic homogeneity of the layers and thus for determining phase separation processes in polar adsorbate assemblies down to inhomogeneities at the molecular level.

Self-assembled monolayers (SAMs) containing terminal polar groups are commonly used for changing the work function of electrode materials.[1−12] This allows manipulating sample work functions, carrier injection barriers, and/or the chemistry of the SAM–ambient interface. The latter then impacts the growth of the active organic layer on top of the SAM-covered electrode.[10,11,13−16] A possible complication is that changing terminal polar groups simultaneously affects all these properties.[17−20] To overcome this problem, systems containing polar units embedded into the molecular backbones are a promising alternative.[21] In this context, we have recently suggested the use of aromatic, terphenyl-methanethiol derived SAMs in which the central ring has been substituted by a 2,5-pyrimidine unit bearing a sizable dipole moment of ∼2.3 D (cf., TP1-up and TP1-down SAMs in Figure a).[22,23] An advantage of these molecules is that the embedded dipoles can be used to control the interface electronic structure without introducing any chemical or sterical constraints. TP1-up and TP1-down form well-ordered layers on Au(111), and depending on the orientation of the pyrimidines, they change the work function of the SAM-covered surfaces by ca. ±0.5 eV compared to a terphenylthiolate-based reference, as shown by Kelvin probe measurements and ultraviolet photoelectron spectroscopy (UPS).[22]

Figure 1

(a) Chemical structures of the TP1, TP1-down, and TP1-up molecules. The orientation of the electric dipole moment of the embedded pyrimidine groups is indicated by blue arrows. (b) Schematic of a TP1 SAM adsorbed on a Au(111) substrate (equivalent to the structures of pure and mixed TP1-up and TP1-down SAMs). The √3 × 3 (short dashed cyan line), 2√3 × 3 (dashed red line), and 6√3 × 3 (solid black line) surface unit cells were used to describe pure films, homogeneous mixtures, and short-range phase separated structures. The blue and green shaded areas indicate the TP1-up and TP1-down arrangement in mixed SAMs. The situation in the striped (inhomogeneous) 50:50 mixture is indicated in the upper half of the figure, whereas the lower half of the figure represents the checkerboard structure of a homogeneous 50:50 mixture. The central molecules of each domain of the striped structure are indicated by yellow ellipses.

Going beyond homogeneous SAMs, mixed monolayers consisting of molecules containing differently oriented dipole moments allow a continuous tuning of the system work function.[24−27] Also for those it is highly desirable to achieve this work function tuning without compromising the properties of the SAM–ambient interface. This has recently been elegantly achieved using mixtures of carboranethiol isomers with varying molecular dipoles.[11] There, the authors not only observed a continuous tunability of the sample work function in mixed monolayers but also achieved that with only very minor changes of the measured water contact angles, surface energies, and thin film growth mode on top of their SAMs.

Here, we focus on mixed monolayers of the embedded dipole molecules TP1-up and TP1-down (vide supra), which have two distinct advantages over the carboranethiols: (i) The dipolar units are even more strongly shielded from the SAM–ambient interface by the terminal unsubstituted phenyl ring (see Figure a). (ii) The X-ray photoelectron spectroscopy (XPS) signal originating from the outermost ring can be used as an efficient tool for probing the distribution of the electrostatic potential within the SAM.[22,28] The latter allows us to directly probe the degree of phase separation within the mixed monolayer (vide infra). This is absolutely crucial for judging the suitability of mixed monolayers for manipulating charge-injection barriers, because phase separation could potentially have highly undesired consequences like the formation of local injection hot-spots. Here we will demonstrate that a combination of XPS and quantum-mechanical simulations enables the determination of the homogeneity of the mixed SAMs at a molecular level.

Mixed TP1-up/TP1-down SAMs were prepared by coadsorption, using the standard immersion procedure and varying the molar ratio of both components in the mixed solutions. SAM-induced work function changes were determined using a Kelvin probe and high-resolution XPS (HRXPS) spectra were recorded at a synchrotron. To simulate the electronic properties of the SAMs, we performed slab-type band-structure calculations on periodic surfaces employing the VASP[29−32] code in conjunction with the program GADGET[33] for geometry optimizations. Simulated XP spectra were obtained employing the approach described in ref (28), with the variance of the individual Gaussian peaks chosen as 0.1 eV to reproduce the experimental peak width of the pure TP1-up SAM. Further details on the experimental and modeling methodology can be found in the Supporting Information.

Consistent with STM experiments on TP1-up, TP1-down, and TP1 monolayers,[22] a (√3 × 3) surface unit cell containing two molecules was chosen for pure monolayers. The molecules were arranged in a herringbone pattern. When homogeneously mixed SAMs were simulated, a (2√3 × 3) unit cell containing four molecules was selected. This allowed the realization of 75:25, 50:50, and 25:75 mixing ratios (see Figure b, bottom half). To simulate short-range phase separation,[34−36] we designed a (6√3 × 3) surface unit cell containing six TP1-up and six TP1-down molecules arranged such that they form homogeneous stripes (see Figure b, top half). Other shapes of short-range phase-separated structures that would require significantly larger unit cells are computationally not tractable. The qualitative conclusions described below, however, also hold for other “shapes” of phase separation, as can be shown by simple electrostatic considerations described in the Supporting Information. Long-range phase separation corresponds to the coexistence of extended TP1-up and TP1-down domains. Therefore, we modeled this situation by an appropriately weighted superposition of the spectra of the neat films for both experimental and simulated spectra.

For mixed SAMs to be of practical use, the first criterion is that varying the mixing ratio allows a tuning of the SAM-induced work function change, ΔΦ. Correspondingly, work function changes as a function of the TP1-up to TP1-down mixing ratio are shown in Figure . In both simulations and experiments we see a continuous evolution of ΔΦ with mixing ratio resulting in an increase of the work function with increasing the TP1-down fraction. In this context it needs to be stressed that the compositions given for the experimental data points in Figure are the ones prevalent in the solutions used to grow the films because direct determination of the film composition after growth was not possible due to the chemically very similar structures of TP1-up and TP1-down.

Figure 2

SAM-induced work function change for a mixed TP1-up:TP1-down SAM as a function of the mixing ratio compared to a TP1 reference SAM. Blue triangles represent average values of two independent experiments (see the Supporting Information for more details); black circles denote simulations. Note that the mixing ratio for the experiments refers to the composition of the solution from which the SAM was grown (which deviates from the actual composition on the surface; see main text for details); in contrast, the mixing ratio in the simulations denotes the composition of the surface unit cell. The blue dashed and the black dash–dotted lines are guides to the eye.

In the simulations the net shifts are somewhat larger than in the experiments, which at least in part can be attributed to structural imperfections like grain-boundaries (cf, discussion in ref (22)). At first glance more surprising are the different shapes of the work function evolutions. In the simulations, an essentially linear dependence on the mixing ratio is obtained in spite of depolarization effects[37−39] prevalent in densely packed SAMs. A similar trend has been found for tail-group substituted SAMs, and its origin is explained in ref (25). In the experiments, a more “S-shaped” curve is observed. That is, even when the mixing ratio in solution deviates from 50:50 (such as 75:25), the absolute magnitudes of the measured work function change appears comparably small. This is not a consequence of depolarization effects as these would result in an opposite curvature. Instead, we attribute it to a film composition on the surface deviating from that in solution, which arises from an increased stability of a 50:50 mixed SAM due to the stabilizing head-to-tail arrangement of neighboring dipoles. This notion is consistent with our simulations, where we find that the total system energy per molecule of the 50:50 mixture is 75 meV lower than the 25:75 up:down mixture and 60 meV lower than the 75:25 mixture. Overall, this behavior is not surprising considering that situations with surface coverage ratios deviating from concentrations in solution have been repeatedly observed in the literature.[11,36,40,41] The symmetric shape of the experimentally observed work function evolution with concentration in solution also implies that there is no preferential deprotonation of either TP1-up or TP1-down on the surface, which can be understood from the fact that the thiolate is well separated from the pyrimidine by a phenyl and a methylene group.

Having established that mixing TP1-up and TP1-down is a viable strategy for changing the work function, it is crucial to determine whether such mixtures are homogeneous or whether significant variations of the local injection barrier are to be expected when using the SAM-covered electrodes in (opto)electronic devices. Our Kelvin-probe measurements are not suitable for that task because they average over large sample areas. Looking for double-peak structures or shoulders in the secondary cutoff region of photoelectron spectra[42] is also not ideal because this would again not reveal inhomogeneities on the nanometer scale.[43] A potential solution would be Kelvin-probe force microscopy measurements, but this technique has also certain limitations;[44,45] therefore, in spite of the recent progress,[46−50] it should be hardly possible to map the work function in SAMs with molecular resolution. Thus, we need a technique that is capable of locally determining the “electrostatic homogeneity” of an adsorbate layer. In the following we will demostrate that HRXPS serves that purpose with a capacity to reveal electrostatic sample inhomogeneities occurring on the scale of ∼10 Å.

As we have discussed recently, the binding energies measured in HRXPS for adsorbate layers not only are determined by the immediate chemical environments of the individual atoms but also crucially depend on the local electrostatic energy.[28] For example, in layers containing ordered assemblies of dipoles, the binding energies for chemically identical species above and below those dipolar sheets can be shifted by several hundred millielectronvolts because of the dipole-induced discontinuities in the electrostatic energy, often also referred to as collective electrostatic effects.[21,22,28] Similar electrostatically induced energetic shifts have been observed for the valence region of polar adsorbate layers.[51] In homogeneous TP1-up and TP1-down SAMs we have measured [calculated] shifts of the C 1s binding energies of the terminal rings amounting to −0.65 eV (TP1-up) and +0.42 eV (TP1-down) [−0.73 eV (TP1-up), + 0.74 eV (TP1-down)] relative to a TP1 SAM not containing any embedded dipoles.[22] As the electrostatic shifts arise from the collective superposition of the fields of the regularly arranged dipoles, this shift has to decrease as soon as the net dipole density decreases, which is exactly what we see in the simulations: for homogeneously mixed SAMs (Figure a), there is (i) a continuous shift of the maximum of the main peak to lower binding energies with increasing TP1-down content. This peak is associated with core-level excitations in carbon atoms of the topmost ring (c.f., ref (22) and calculated core-level binding energies contained in the Supporting Information). (ii) The full width at half-maximum (fwhm) of the main peak does not change; and (iii) the peak positions for a 50:50 mixed TP1-up:TP1-down SAM and a TP1 SAM are very close (cf., Figure a,b).

Figure 3

Normalized calculated XP spectra of pure and mixed TP1-up:TP1-down SAMs. (a) Calculated XP spectra for pure TP1-up (black) and TP1-down (cyan) SAMs as well as homogeneously mixed SAMs [at TP1-up to TP1-down ratios of 75:25 (red), 50:50 (green), and 25:75 (blue)]. (b) Calculated spectra for a TP1 reference SAM (dashed violet), the 50:50 striped (inhomogeneous) mixture (dark green), and the simulated spectrum of coexisting large TP1-up and TP1-down domains (gray) obtained by a weighted superposition of the spectra of the pure TP1-up and TP1-down SAMs. The orange curve displays the spectrum of the striped (inhomogeneous) SAM, which was calculated considering only the two central molecules of each domain to reduce edge effects (molecules highlighted by yellow ellipses in Figure b). All spectra of panels a and b were obtained using a variance of 0.1 eV to match experimental peak widths. (c) XP spectra of the striped SAM (dark green and orange) and the TP1 reference SAM (dashed violet) evaluated with a reduced variance of 0.03 eV to better visualize the origin of the broadening.

The situation changes significantly as soon as phase separation effects are considered (Figure b,c). For the striped structure at 50:50 mixing ratio representing short-range phase separation, a significant increase of the fwhm of the main peak compared to the homogeneous mixture is observed (i.e., the fwhm becomes 1.17 eV instead of 0.83 eV). As the obtained spectrum is dominated by molecules at the edges of the domains, it is useful to consider also the situation of the central TP1-up and TP1-down molecules in the stripes (highlighted by yellow ellipses in Figure b; orange curve in Figure b). This corresponds to disregarding the most pronounced edge effects. In that case, we obtain a further broadening of the main peak to a fwhm of 1.46 eV, where the maximum is rather a plateau indicating the onset of a double-peak feature. The latter is better resolved when reducing the extrinsic peak width, as shown in Figure c. The plot at reduced extrinsic broadening reveals that energetically distinct peaks are also responsible for the broadening when considering all molecules of the striped structure (see also calculated core-level energies in the Supporting Information).

The situation becomes most extreme when considering long-range phase separation representing extended TP1-up and TP1-down islands in which molecules are surrounded essentially only by identical neighbors and electrostatic shifts become a maximum (gray curve in Figure b) resulting in two clearly distinct peaks separated by 1.53 eV. The above considerations show that even very short ranged inhomogeneities of the films result in a broadening of the spectra as different molecules experience different electrostatic environments. Here we take advantage of the fact that XPS allows us to very locally probe the electrostatic environment of certain atoms within a molecule. An in-depth discussion of the shapes of the XPS spectra can be found in the Supporting Information.

The different electrostatic environments for the top rings become apparent in Figure a, where the electrostatic energy of the electrons in the plane of half of the molecules of the unit is shown for the striped phase. The differences are best resolved (by the coloring and isodensity lines) in the intermolecular regions left and right of the central molecules of the stripes (marked by arrows). No such variations are visible for the homogeneously mixed SAM shown in Figure b.

Figure 4

Electrostatic energy of an electron plotted for the striped (inhomogeneous) 50:50 SAM (top panel) and the homogeneous 50:50 SAM (bottom panel) in the plane containing the long molecular axes of half of the molecules (aligned along the x-axis of the unit cell). Values are given relative to the Fermi level of each system. The black isodensity lines cover the energy interval from 3.5 to 4.3 eV (the same interval in which the colors are varied) with lines drawn every 0.05 eV. To depict the much stronger energy variations inside the SAM, red isodensity lines are plotted, covering the range from 4.3 to 6.8 eV with a spacing of 0.25 eV.

What is even more striking when comparing these two plots are the fundamentally different energy landscapes right above the SAMs. While the electrostatic energy is essentially constant directly above the homogeneously mixed monolayer (Figure b), there are non-negligible potential variations above the SAM in the striped phase extending several angstroms into the region above the monolayer. This is relevant insofar as the potential variations can be expected to increase for larger domain sizes. Moreover, bearing in mind the exponential dependence of thermionic and tunneling injection on barrier heights, they will have a non-negligible impact on local charge carrier injection efficiencies, even when screening in a subsequently adsorbed semiconductor layer might somewhat reduce the effect.

To identify the situation prevalent in the experimental samples, we finally compare the simulated XPS spectra to the results of the HRXPS investigations on the mixed SAMs. These are shown in Figure a. They portray a situation fully consistent with the theoretical predictions for homogeneously mixed SAMs: (i) There is a continuous shift of the peak maximum with mixing ratio. (ii) There are no traces of a double-peak structure. The experimental spectra also bear no resemblance to the hypothetical spectra for the coexistence of large domains (Figure b) that have been obtained by a weighted superposition of the spectra measured for the single-component TP1-up and TP1-down films (i.e., the experimental equivalents to the gray curve in Figure b). (iii) Finally, no peak broadening is observed in the mixed SAMs (cf., Table S2 containing the fwhms of the main peaks of the simulated and measured spectra presented in the Supporting Information).

Figure 5

(a) Experimental C 1s HRXPS spectra of the homogeneous TP1-up (blue), TP1-down (red), and mixed TP1-up/TP1-down (black) monolayers acquired at a photon energy of 350 eV and (b) weighted sums of the TP1-up and TP1-down spectra, corresponding to a coexistence of large domains in the mixed films. The percentages of the TP1-up and TP1-down molecules in the solutions are color-coded and given at the respective spectra. The vertical olive lines in panel a mark the positions of the dominant emission. The thick horizontal bars between these lines mark the energy shifts. The overall rigid shift between the experimental and simulated (Figure ) spectra is a consequence of the initial state method of calculating core level energies and is not relevant for the present comparison.[52−55]

Interestingly, the continuous shift of the peak maxima in Figure a mimics the behavior of the “S-shaped” work function curve in Figure . I.e., it is comparably small for the 25:75 and 75:25 up:down mixtures relative to the 50:50 composition and significantly larger between the 25:75 and 75:25 mixtures and the respective pure films (mixing rations in experiments again referring to the situation in the solutions used to grow the films). Bearing in mind that the simulations predict a homogeneous shift of the peak maximum at the varying film composition (Figure a), this once again points toward a different composition of the films on the surface compared to the situation in solution with the 50:50 mixture being energetically favored.

Note that along with the dominant peak addressed above, the spectra in Figure exhibit some further minor peaks. These are the features at ∼284.8 and 285.6 eV. They are associated with the pyrimidine moiety which are best perceptible in the spectrum of the pure TP1-down SAM as well as the feature at ∼283.85 eV associated with the substrate-adjacent phenyl ring which is best seen in the spectrum of the pure TP1-up SAM (see the Supporting Information and especially ref (22) for details). These features decrease in intensity and overlap with other peaks upon mixing with either TP1-up or TP1-down.

In conclusion, we have shown that mixed monolayers consisting of TP1-up and TP1-down molecules do not only allow a continuous tuning of the sample work function in the far-field, i.e., when measuring work function changes using a Kelvin probe. Much more importantly, an in-depth analysis of the obtained high-resolution XPS spectra in conjunction with suitable DFT-based simulations indicates that phase separation is essentially absent in the films. This results in a very smooth electrostatic energy already in the immediate vicinity of the SAMs, which is a prerequisite for using such mixed SAMs for controlling the injection barrier of electrodes in (opto)electronic devices.